USING NEXT-GEN SEQUENCING TO MAP UG99 STEM RUST RESISTANCE GENES IN WHEAT BIPARENTAL AND NESTED ASSOCIATION MAPPING POPULATIONS

USING NEXT-GEN SEQUENCING TO MAP UG99 STEM RUST RESISTANCE GENES IN WHEAT BIPARENTAL AND NESTED ASSOCIATION MAPPING POPULATIONS A DISSERTATION SUBMIT...
Author: Joanna Fisher
22 downloads 2 Views 1MB Size
USING NEXT-GEN SEQUENCING TO MAP UG99 STEM RUST RESISTANCE GENES IN WHEAT BIPARENTAL AND NESTED ASSOCIATION MAPPING POPULATIONS

A DISSERTATION SUBMITTED TO THE FACULTY OF UNIVERSITY OF MINNESOTA BY

PRABIN BAJGAIN

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

JAMES A. ANDERSON, ADVISER

October 2014

© Prabin Bajgain 2014

Acknowledgements I would like to express my sincere appreciation towards everyone who assisted me with the projects that make up this thesis. I especially thank my adviser Dr James Anderson for his guidance, mentorship, and support throughout these years. I greatly appreciate his constant care and concern towards my personal as well as professional progress; and am forever indebted for the opportunity he provided me to work with him. I wish to extend my gratitude to the members of my graduate research committee, Dr Brian Steffenson, Dr Kevin Smith, Dr Matthew Rouse, and Dr Robert Stupar for providing thoughtful suggestions regarding my research. I am particularly grateful towards Dr Matthew Rouse for coordinating the planting of my materials in East African stem rust nurseries, and for recording phenotypic data when I was unable to travel. I also thank the members of the Wheat Breeding Project, Brian Seda, Kathryn Turner, Emily Conley, Catherine Springer, Jen Flor, Susan Reynolds, Roger Caspers, Dr Godwin Macharia, Dr Itamar Nava, Dr Liangliang Gao, and Dr Xiaofei Zhang and all undergraduate students (especially Brittney Johnson) associated with the breeding project from Fall 2011 to 2014. Their technical assistance and advice, as well as the friendship have been highly valuable in completing my research work. Thanks to Justin Anderson, Lian Lian, Patrick Ewing, and Ahmad Sallam for being excellent officemates, and to fellow APS graduate students for many fruitful discussions and the camaraderie. I thank the faculty and staff in the Department of

i

Agronomy and Plant Genetics, whose academic knowledge, and research expertise were of great resources to me. Lastly, I thank the Cereal Disease Laboratory, the University of Minnesota Genomics Center, Minnesota Supercomputing Institute, the Microbial & Plant Genomics Institute, the University of Minnesota Graduate School, Kenya Agricultural Research Institute, Ethiopia Institute of Agricultural Research, and the Agricultural Research Council of South Africa for their help and support during various phases of the study. Funding for most of the work discussed herein was provided by the United States Department of Agriculture, Agriculture and Food Research Initiative and 2011-6800230029 (Triticeae Coordinated Agricultural Project, www.triticeaecap.org), and the Borlaug Global Rust Initiative Durable Rust Resistance in Wheat Project (administered by Cornell University with a grant from the Bill & Melinda Gates Foundation, and the UK Department for International Development).

ii

Dedication I dedicate this thesis, with much humility and love, to my parents, Prabha and Purushotam, who while living half the circumference of the earth away, always believed in me, and supported my endeavors. Apart from the encouragement they provided to pursue the highest academic degree possible, they have inspired me to always be patient and humble; and practice kindness and compassion with all living beings, which I believe, define a person more so than just formal education.

iii

Table of Contents Acknowledgements

i

Dedication

iii

Table of Contents

iv

List of Tables

vii

List of Figures

ix

List of Appendices

xi

Chapter 1

1

A Historical Account of Resistance Breeding and Genetic Mapping of Stem Rust Resistance in Wheat

1

Stem Rust Epidemics

1

The Emergence of Ug99

4

Gene Discovery & Resistance Breeding

6

Types of Resistance and Disease Control

8

Stem Rust Resistance Gene Mapping

10

Gene Mapping Tools & Strategies

15

Chapter 2

17

Mapping Putatively Novel QTL Conferring Adult Plant Resistance to Ug99 in the Biparental Population RB07/MN06113-8

17

Introduction

19

Materials and Methods

23

Plant Materials

23

Field Stem Rust Evaluation

24

Phenotyping and Data Analysis

25 iv

Molecular Marker Assay

26

Linkage Map Construction and QTL Mapping

28

Results

30

Disease Evaluation

30

Construction of Linkage Maps of the RIL Population

31

Quantitative Mapping of Resistance to Stem Rust

31

Discussion

33

Conclusion

39

Chapter 3

49

Comparison Between SNP-chip based Genotyping and Genotyping-by-Sequencing Approaches in Linkage Mapping of Quantitative Trait Loci in Wheat

49

Introduction

51

Materials and Methods

55

Molecular Marker Assay

55

Linkage Map Construction and QTL Mapping

56

Imputation of GBS SNPs

57

Summary Statistics of Genotype Matrices

58

Methodology and Workflow Comparison

59

Results and Discussion

61

Genotype Properties

61

Linkage Groups Construction

61

Recombination and Genome Coverage

63

QTL Mapping

66

Imputation of GBS Markers

67

QTL Mapping Using Imputed Datasets

69

Comparison of Methodology and Required Resources

70

Conclusion

74 v

83

Chapter 4 Nested Association Mapping of Stem Rust Resistance in Wheat Using Genotyping by Sequencing

83

Introduction

85

Materials and Methods

90

Plant Material

90

Disease Phenotyping

90

Statistical Analysis

92

Genotyping, SNP Discovery & Imputation of Missing Alleles

93

Map Construction and Joint QTL Mapping

96

Results and Discussion

97

Genotyping

97

SNP Discovery and Linkage Mapping

98

Population Characteristics and Stem Rust Reaction

99

Stem Rust QTL Mapping

101

Comparison of Joint Mapping to Single-population QTL Mapping

104

Comparison of Joint Mapping Results to Previously Reported Genes and QTL Conferring Resistance to Pgt

107

Conclusion

113

References

121

Appendix

140

vi

List of Tables Chapter 2 Table 1: Mean values and range of stem rust severity (%) in the RIL mapping population derived from RB07/MN06113-8 in four field environments at Njoro, Kenya, Debre Zeit, Ethiopia, and St. Paul, MN, USA.

Table 2: Pearson correlation coefficients of stem rust severity observed in the RB07/MN06113-8 population in four field environments at Njoro, Kenya, Debre Zeit, Ethiopia, and St. Paul, MN, USA.

40

41

Table 3: Quantitative trait loci (QTL) for APR to stem rust detected in the RB07/MN06113-8 population of 141 recombinant inbred lines by composite interval mapping in four environments. 42 Table 4: Reduced stem rust severity in the RB07/MN06113-8 RIL population by combinations of QTL n pairs in each environment.

44

Chapter 3

Table 1: Results of linkage groups formation using the 9K and GBS SNPs.

75

Table 2: Comparison of linkage mapping results among the 9K dataset, non-imputed GBS dataset, and datasets with 40% and 75% missing allele calls.

76

Chapter 4 Table 1: Origin, pedigree, and stem rust reaction of parent lines used to develop the NAM population.

vii

114

Table 2: Common quantitative trait loci (QTL) detected between the iterative QTL mapping (iQTLm) in nested association mapping of 10 RIL populations, and composite interval mapping (CIM) methods in individual populations for stem rust adult plant resistance in four environments.

viii

116

List of Figures Chapter 2 Figure 1: Frequency distribution of stem rust severity (%) for the RB07/MN06113-8 population comprised of 141 recombinant inbred lines evaluated in four environments – two at Njoro, Kenya, one at Debre Zeit, Ethiopia, and one at St. Paul, MN, USA.

40

Figure 2: Distribution of markers (both 9K SNPs and GBS de novo SNPs) across the 21 wheat chromosomes. 41 Figure 3: Quantitative trait loci (QTL) interval map for QSr.umn-2B.2 detected on wheat chromosome 2B. 42

Chapter 3 Figure 1: Recombination blocks observed in 9K (Panel A) and GBS (Panel B) genotype datasets; and a representative example of the difference in genome sampling between the 9K and GBS methods (Panel C). 79 Figure 2: Characteristics of genotype matrices in imputed and non-imputed datasets.

80

Figure 3: A comparison of the QTL detected on chromosome 2B among the three GBS datasets: non-imputed GBS dataset (Panel A), GBS40 (Panel B), and GBS75 (Panel C).

82

Chapter 4 Figure 1: Schematic workflow of the study. See Materials & Methods for detailed explanation of each step.

117

Figure 2: Distribution of SNPs by linkage groups (X-axis) and read depth (Y-axis) obtained from GBS approach.

118

ix

Figure 3: Heat map of additive effect estimates of alleles contributed by the 10 founder lines at the QTL for resistance to Pgt races.

119

Figure 4: Frequency distribution of number of QTL discovered in different environments from composite interval mapping (CIM) of each RIL population and iterative QTL mapping (iQTLm) in joint mapping of all populations combined.

120

x

List of Appendices Appendix I: QTL detected via CIM in the 9K (Table A), non-imputed GBS (Table B), GBS40 (Table C), and GBS75 (Table D) datasets.

140

Appendix II: Quantitative trait loci (QTL) detected for stem rust adult plant resistance by joint mapping of 10 RIL populations in four environments. 143 Appendix III: QTL detected via CIM in the RIL populations LMPG-6/Ada (Table A), LMPG-6/Fahari (Table B), LMPG-6/Gem (Table C), LMPG-6/Kudu (Table D), LMPG-6/Kulungu (Table E), LMPG-6/Paka (Table F), LMPG-6/Pasa (Table G), LMPG-6/Popo (Table H), and LMPG-6/Romany (Table I). 146 Appendix IV: Distribution of sequence reads by each population and parent (scaled down to 1/10) generated using the GBS approach. 149 Appendix V: A scatter plot of principal component 1 (PC1) plotted against principal component 2 (PC2) of the NAM population. 150 Appendix VI: Frequency distribution of stem rust severity (%) for the ten NAM populations in four environments – St. Paul 2012 (StP12), South Africa 2012 (SA12), St. Paul 2013 (StP13), Kenya 2013 (Ken13).

151

Appendix VII: Quantitative trait loci (QTL) detected in 11 chromosomes using the iQTLm method of joint mapping approach in four environments.

159

Appendix VIII: Field reaction to stem rust observed on the RB07/MN06113-8 RIL population. 163 Appendix IX: Field reaction to stem rust observed on the 10 NAM RIL populations. 167

xi

Chapter 1

A Historical Account of Resistance Breeding and Genetic Mapping of Stem Rust Resistance in Wheat

Records indicate that the documentation of stem rust of wheat (caused by Puccinia graminis Pers. f. sp. tritici) likely began with simple field observations, and can reasonably be traced back to the Late Bronze Age II (1400 – 1200 B.C.) (Arthur et al. 1929; Kislev 1982). From field observations in ancient Rome (Fantham 1998) and ancient Greece (Theophrastus) to taxonomical classification of the pathogen (Tulasne and Tulasne 1847; Tulasne 1853; Persoon 1794) and dissection of its life cycle (de Bary 1853), the disease stem rust of wheat as well as the pathogen causing the disease have been thoroughly studied. Rich historical data collected have paved the way for modern epidemiologic and genetic studies, which continue to add to the voluminous nature of research work carried out on this disease. Stem Rust Epidemics The destructive nature of stem rust on cultivated wheat has been recorded during epidemics throughout both ancient and modern agricultural periods. The first documented stem rust epidemic came from Europe in 1767 when two Italian scientists, Felice Fontana (1767) and Giovanni Targioni Tozzetti (1767), independently published their observations. Both authors stated that stem rust was prevalent in all wheat growing 1

regions of Italy, indicative of an epidemic. Other major rounds of severe epidemics in Europe occurred in 1932 in central Europe with recorded yield losses of up to 20%; and in 1951 in Scandinavia, with losses ranging 9-33% (Zadoks 1963). No major outbreaks have been reported in Europe after these reports. Nearly around the same time the stem rust epidemic was observed in Italy in 1767, wheat stem rust may have also been a big problem in North America (Roelfs 1985a). Wheat rusts were observed in the United States as early as 1660 (Bidwell and Falconer 1925; Fulling 1943). Despite the lack of written evidence of a stem rust epidemic, Alan Roelfs hypothesizes that passing of barberry eradication laws in the Eastern United States in the 1750s and 1760s suggests localized outbreaks of stem rust in those areas. Fulling (1943) states that the wheat production losses in the Eastern United States was connected to the presence of barberry bushes, which led the states of Connecticut, Massachusetts, and Rhode Island to pass barberry eradication laws in 1726, 1764, and 1766, respectively. The first recorded stem rust epidemic in the United States was observed in 1878, affecting Northern Iowa, Southern Minnesota, and Wisconsin (Hamilton 1939). These areas experienced severe yield losses from additional wheat stem rust epidemics in 1904 (Carleton 1905), 1916, 1923, 1925, 1935, 1937, and 1953-1954 (Stakman and Harrar 1957). Of all these epidemics, the 1916 outbreak holds a significant value because of the level of awareness it raised towards resistance breeding and disease management. It was after the 1916 epidemics that barberry eradication laws were passed and implemented on the national scale in an effort to keep the pathogen population in check. This epidemic also led to identification of sources of resistance, and their 2

introgression into hard red spring bread wheat lines that were later released as varieties (McFadden 1930). Several rust epidemics in Canada coincided with the US outbreaks. Canadian wheat growing regions observed rust epidemics mainly in 1904, 1909, 1916, 1919, 1923, 1925, and 1927. The rust outbreaks of the mid-1930s and mid-1950s in the United States are attributed to evolution of new pathogen races, with race 56 being prominent in the 19351937 epidemics, and race 15B in the 1953-1955 epidemics. No major stem rust outbreak was observed between 1938 and 1952 due to control of the pathogen population aided by barberry eradication in addition to use of resistant cultivars. There have been no widespread stem rust epidemics in the Northern Great Plains after the last epidemic observed during the 1950s, though localized epidemics have been occasionally observed in the southern states (Roelfs 1985a; Brian Steffenson, personal communication). In the past decade and a half, a highly virulent race TTTTF has been observed in North America (Jin 2005). However, this pathogen race does not pose an imminent risk to wheat crop mainly because the resistance genes Sr24 and Sr31, along with other unknown genes safeguard wheat crops from this race (Olson et al. 2010b; Klindworth et al. 2011). In addition to North American and European epidemics, several other wheat growing regions of the world also have experienced stem rust outbreaks, albeit less frequently and of less damaging nature. Stem rust epidemics in Australia have been observed in 1889 (McAlpine 1906), 1903, 1916 (Waterhouse 1929), 1947 (Butler 1948), and 1974 (Watson 1981), causing severe yield loss. Similarly, epidemics have also been observed in China in 1948, 1951-1952 and 1956 (Roelfs 1977), and intermittently and 3

less severe, only during unusually warm seasons in India (Joshi and Palmer 1973). The outbreaks in Australia and China are credited mainly to an unusual rise in temperature during those years. Pathogen race survey and identification projects in East Africa suggested that small, scattered wheat growing areas are prone to the attack of the pathogen (Green et al. 1969; Harder et al. 1971). In Kenya, stem rust epidemics might have been observed as early as in the 1910s and 1920s, which led to establishment of rigorous resistance breeding programs (Burton 1928; McDonald 1931). In summary, the historical data suggest that stem rust of wheat has been a major problem in most wheat growing regions of the world. The data also document effective control of the disease by using a combination of resistance genes and other protective measures such as fungicide application, with the former being the preferred and most effective method. We can project that this disease can be effectively controlled if sources of resistance are used timely and effectively, with regular monitoring of evolution and spread of the pathogen. The latter seems to be of high importance, as manifested by the evolution and dominance of the Ug99 group of races in Africa. The Emergence of Ug99 After decades of relative absence of epidemic proportions of the disease came the untimely reminder that this historical foe of the wheat crop is not quite ready to give up. An evolved and much more damaging isolate of the pathogen was discovered in Uganda in 1998, and was named Ug99 for the country of origin and the year it was evaluated for its virulence phenotyping (Pretorius et al. 2000). Later, the isolate was designated as race TTKS according to the North American race nomenclature system (Wanyera et al. 2006). 4

With the addition of four more lines to the standard 16 wheat stem rust differential lines set, race TTKS was subsequently keyed to TTKSK (Jin et al. 2008). At its discovery, isolate Ug99 had overcome the most important stem rust resistance gene of the time: Sr31. The gene Sr31 was transferred to hexaploid wheat from the short arm of chromosome 1 of rye (Secale cereale) (Zeller 1973). This gene had been widely deployed by CIMMYT (The International Maize and Wheat Improvement Center, Mexico) because of its high level of resistance to several diseases, broad adaptation, and higher yield potential (Dubin and Brennan 2009). Sr31 remained highly effective and durable globally for more than three decades until isolate Ug99 arose. Within a few years of its discovery, variants in the Ug99 lineage were detected that carried virulence to lines with genes Sr24 and Sr36 in Kenya in 2006 and 2007, respectively (Jin et al. 2008; Jin et al. 2009b). Originally detected as a single race in East Africa, isolate Ug99 has now evolved to include seven races (TTKSK, TTKST, TTTSK, TTKSF, TTKSP, PTKSK, and PTKST) in its lineage, with demonstrated virulence on a number of important stem rust genes (Singh et al. 2011). These races have spread, as documented by their arrival in Iran in the north and South Africa in the south. The races not only travel long distances, but also evolve further in regions of deposition (Singh et al. 2011; Pretorius et al. 2012). However, the most disturbing aspect of this movement and evolution is that most of the existing wheat varieties and germplasm are susceptible to the Ug99 race group. During 2005-2010, more than 200,000 wheat accessions representing advanced breeding materials from several wheat breeding programs around the world, and germplasm collections were screened in Kenya and Ethiopia (Singh et al. 2006; Singh et al. 2008b). 5

The results showed that up to 95% of the tested materials are susceptible to the races. This is a terrifying scenario, especially considering that these new virulent African stem rust races are not yet completely under control, and are capable of traveling long distances. It might only be a matter of time before these races make into the bread baskets of Asia, Europe, and the Americas. In an effort to minimize future stem rust losses, the Durable Rust Resistance in Wheat (DRRW) project was funded as a part of the Borlaug Global Rust Initiative (BGRI). BGRI is an international collaborative community that aims to reduce the world’s vulnerability to all three rust diseases of wheat (leaf, stem, and stripe) by facilitating an international system to contain their threat. Because of the initiative taken by BGRI as well as all collaborating wheat breeding programs in several countries, breeders and pathologists worldwide are now better prepared to face the disease. However, the competition between rust and the host is a constant arms’ race, and as such, continual discovery and deployment of resistance genes is needed. Gene Discovery & Resistance Breeding Records indicate that wheat varieties resistant to the rusts were in cultivation as early as 1841 (Chester et al. 1951); however, systematic breeding for resistance and gene deployment did not become common until the early 1900s (Roelfs 1985a). The credit for identification of the earliest source of resistance to wheat stem rust goes to Herbert K. Hayes (1925). He crossed the hard red spring line ‘Marquis’ with Iumillo durum, and progenies of this cross protected wheat planted in the fields of the northern United States from leaf and stem rust. ‘Thatcher’, one of the cultivars developed from this particular cross, was a popular cultivar and has served as the source for many hard red spring wheat 6

lines (Roelfs 1985a). The first documented introgression and deployment of a stem rust resistance gene is that of Sr2 (McFadden 1930). In the 1920s, Edgar S. McFadden made selections in the field from the cross he made between emmer wheat (T. turgidum var. dicoccum cv. Yaroslav) and a hexaploid wheat ‘Marquis’. He labeled his selections as varieties ‘Hope’ and ‘H44-24’ (also known as ‘H-44’), and these two sib lines are still being used worldwide to transfer Sr2 resistance to modern germplasm. The gene Sr2 has been highly durable and serves as a landmark gene by continuing to provide partial, adult plant resistance to all known races of stem rust. The work of gene discovery and deployment that began in the 1920s has been religiously continued, with 58 stem rust resistance genes being discovered and documented to date (McIntosh et al. 1995; McIntosh et al. 2003; McIntosh et al. 2011). Most of the genes catalogued in these documents have been contributed by US or Canadian scientists. This is not surprising, as the high frequency of stem rust epidemics observed in North America called for the identification of novel sources of resistance that could protect the wheat crop from newly emerged rust pathogen races. Some of the genes discovered in North America made it to the field quicker, and helped in reducing the damage done by the frequent stem rust epidemics observed in these areas. However, many of the deployed genes had to be replaced within a few years as the selection pressure on pathogens led to evolution of new, virulent rust pathogen races that overcame the deployed genes. This cyclical process of gene discovery, deployment, and defeat by a more evolved pathogen, known famously as the ‘boom and bust cycle’ (Browning 1979), can be reviewed in a few case studies in North America. For example, the gene Sr9d was 7

used in bread wheat and durum wheat in North America, and was resistant to race 56 during the 1934-1937 epidemics. However, race 15B, which arose to cause the 1950s epidemics overcame Sr9d (McIntosh et al. 1995). The genes Sr12 and Sr14 also bear a similar story during the rust epidemics in North America. In fact, the boom and bust cycle continues today with the emergence of the Ug99 race group. Types of Resistance and Disease Control The preferred approach to combating rust pathogen races has been via the deployment of resistant cultivars. The resistance deployed against stem rust can be grouped into two broad categories: all-stage (or seedling) resistance and adult plant resistance (APR). All-stage resistance is often associated with its hypersensitive reaction upon fungal attack, and confers a high level of resistance throughout all stages of plant growth (Roelfs et al. 1992). Resistance of this type is relatively easy to phenotype in field and greenhouse settings because of the large effect on disease reduction, and have been well catalogued (McIntosh et al. 1995). While all-stage resistance can be bred into desired germplasm with higher degrees of success than APR, the major drawback of using this resistance is the possibility of the pathogen evolving to overcome the deployed resistance. As an alternative, APR, which is generally expressed during the adult growth stages of the plant, has been proposed for durable resistance. APR is sometimes also explained as race-nonspecific resistance or partial resistance, though anomalies to this definition have also been observed in the field. In the wheat rust pathosystems, partial resistance is often called slow rusting resistance, defined by Caldwell (1968) as the type of resistance that would either 1) exclude the fungus, 2) limit pustule size without 8

expressing hypersensitivity, or 3) reduce the overall rate of growth and development of the pathogen fungus. Parlevliet (1976) characterized slow leaf rusting in barley, and explained the observed resistance as a form of incomplete resistance exhibiting a reduced rate of disease development despite the plants developing a susceptible infection type. APR genes are considered more durable than all-stage resistance, with the stem rust resistance gene Sr2 being durable against wheat stem rust for almost nine decades. However, phenotyping and breeding of APR genes that confer partial resistance are difficult, and provide low, generally inadequate, levels of resistance when deployed as single genes. Therefore, deployment of lines with both types of resistance genes pyramided is a more common and desired approach (Singh 2012; Evanega et al. 2014). The disease can also be effectively controlled by the use of fungicides (Cook et al. 1999; Wanyera et al. 2009; Zhensheng et al. 2010). Foliar fungicides have been documented to reduce infection severity on plants, increase grain density as well as grain yield (Mayfield 1985; Wanyera et al. 2009). However, they may come with severe environmental burdens (Steffenson et al. 2007). High costs associated with fungicides may keep them out of reach of smallholder farmers in developing nations. Effective fungicide application largely relies on weather conditions. For example, if there is a rainfall not too long after fungicide application, another round of application might be necessary. Another concern is the development of fungicide tolerance in the pathogen, which render the chemical ineffective for rust control (Gill et al. 1985). Considering these drawbacks, discovery and mapping of genes followed by their introgression into elite

9

germplasm and careful deployment is the ideal strategy for sustainable and durable protection against stem rust. Stem Rust Resistance Gene Mapping While the discovery of resistance genes began in the 1920s, assigning the genes to chromosomes did not commence until the mid 1950s. Even so, this work was limited only to identification of genes in monosomic lines in a Chinese Spring background developed by Ernest Sears (Sears 1954). Seedlings of monosomic lines were inoculated with stem rust races, and genes were mapped (a more appropriate term may be chromosome tagging) to chromosomes through aneuploid analysis. Sr6, a dominant gene, and Sr8, a recessive gene, were among the first stem rust resistance genes assigned to a chromosome (Sears et al. 1957; Sawhney et al. 1981). The seedlings of monosomic Chinese Spring lines were inoculated with stem rust races 15C and 122 to determine the locations of Sr6 and Sr8, respectively. Analysis of segregants revealed the location of Sr6 to be on chromosome 2D, and that of Sr8 on 6A. The development of monosomic lines enabled chromosome tagging of several other genes, namely Sr5, Sr7, Sr9, Sr11, Sr15, and Sr16, carried out by Plessers (1954) and Sears et al. (1957). Mapping stem rust resistance genes solely using aneuploid stocks continued until the early-1990s, where Sr33 (Jones et al. 1991) and Sr26 (Bariana and McIntosh 1993) were mapped using recombinant substitution lines obtained from crossing a double-ditelosomic line with a normal chromosome (1D), and a monosomic series of lines, respectively. A point to note here is that tagging chromosomes (using aneuploids) with genes that were discovered before DNA markers, and marker-based genetic mapping of the same genes have not 10

been a coherent process. In other words, genes discovered before the 1980s were named in the order they were discovered, but their genetic mapping with molecular markers has not always been carried out in the same order of gene discovery. In fact, many of the discovered genes remain unmapped to this day despite their significant contribution to resistance breeding and study of genetic resistance. This is exemplified by the use of several unmapped yet important stem rust genes in the first four sets of differentials used in the North American nomenclature system for race identification of stem rust isolates (Roelfs and Martens 1988). Once the era of molecular markers began with the use of restriction fragment length polymorphism (RFLP) technique for genetic profiling in the late 1980s, accurate genetic mapping of causative loci was possible. Because of the availability of aneuploid lines for each chromosome in hexaploid wheat (Sears 1954), construction of genetic maps of each chromosome to isolate regions of interest was the logical approach. This approach was first implemented by Chao et al. (1989) who reported the construction of RFLP-based linkage map of all three subgenomes of wheat chromosome 7. A series of studies that followed assigned RFLPs to several wheat chromosomal arms (Anderson et al. 1992; Werner et al. 1992; Devey and Hart 1993; Deynze et al. 1995). Information from these studies would be utilized by studies in subsequent years to accurately map stem rust resistance genes, among other genes of interest, to wheat chromosomes. One of the first studies to do so was carried out by Paull et al. (1994) who identified RFLP markers associated with the stem rust resistance gene Sr22 on chromosome 7A. The gene Sr22 is derived from einkorn wheat (Triticum monococcum), and is one of the very few 11

genes that are effective to races in the Ug99 lineage (Kerber and Dyck 1973; Singh et al. 2011). In their study, Paull et al. crossed einkorn wheat T. boeoticum with the bread wheat variety ‘Schomburgk’ to generate F2 lines. Using RFLP markers, the researchers were able to identify lines with low levels of recombination between chromosome 7A of cultivated and chromosome 7Am of einkorn wheat. The study also investigated the amount of linkage drag in the progeny, and demonstrated reduced alien introgression in progeny. Linkage drag leads to fitness reduction when undesired genes get introduced along with beneficial genes during gene introgression from wild species. Nelson et al. (1995) identified the locus Xbcd1095 on the long arm of chromosome 2B that provided major resistance against stem rust of wheat in field conditions. This study hypothesized that the locus could be linked to Sr16, a resistance gene possibly contributed by the line ‘Thatcher’ in the population under study. Shortly after, Sr2 was mapped to chromosome 3B by Bariana et al. (1998) using short tag sequence (STS) markers and also by Johnston et al. (1998) using RFLP markers in an RIL population of 141 lines. Because of the high importance of the gene Sr2, multiple attempts have been made to discover tightly linked markers (Spielmeyer et al. 2003; Hayden et al. 2004; Mago et al. 2011a). Spielmeyer et al. (2003) reported that Xgwm533 was tightly linked with Sr2 (at an approximate distance of 2cM) and produced a 120 bp product in PCR. Later, Mago et al. (2011a) reported an improved marker csSr2 which had an accuracy of 95%, compared to 84% of the marker Xgwm533. Mapping of several other resistance genes that are effective against the Ug99 race group have also been accomplished, in addition to mapping of Sr2 and Sr22. Sr13 is 12

resistant against races TTKSK, TTKST, and TTTSK, and was mapped in tetraploid durum wheat (Triticum turgidum ssp. durum L.) (Simons et al. 2011). Sr47, a gene transferred from Aegilops speltoides to durum wheat, confers high level of resistance to Ug99 stem rust and was mapped by Klindworth et al. (2012). However, markers for both Sr13 and Sr47 are not diagnostic in all genetic backgrounds. Diagnostic markers for the Ug99-effective genes Sr24, Sr25, and Sr26 have been developed (Mago et al. 2005; Liu et al. 2010). Mapping and marker development for the genes Sr28 (Rouse et al. 2012), Sr39 (Mago et al. 2009), and Sr45 (Periyannan et al. 2014) have also carried out, but the markers liked to these genes are not reported to be diagnostic. To date, only two stem rust resistance genes that are also effective against the Ug99 race group have been cloned: Sr33, derived from Aegilops tauschii, was cloned by Periyannan et al. (2013); and Sr35, derived from Triticum monococcum, was cloned by Saintenac et al. (2013b). Despite one of the goals of QTL mapping being map based cloning of the gene of interest, it is not an easy task due to environmental conditions confounding QTL detection, and lack of completely linked markers detected during QTL mapping (Remington et al. 2001; Salvi and Tuberosa 2005). In addition, as one of the culturally most important crops in the world, wheat faces political and behavioral challenges in the areas of cloning and gene transformation. Therefore, candidate gene discovery and recurrent selection using diagnostic molecular markers are still the best means of obtaining superior varieties with enhanced trait values. In addition to the mapped genes, several uncharacterized sources of resistance in common wheat and its diploid progenitors and wild relatives were also documented as 13

early as 1970 (Roelfs and McVey 1979). These resistant sources have not been assigned a gene number because genetic information such as chromosome location and/or tests of allelism with previously designated genes are required to confirm the identity of the gene. This problem exists mainly because gene discovery and mapping have not always been coherent, as mentioned earlier. One method to determine the identity of a gene is by mapping the segregating resistance, commonly known as gene or quantitative trait loci (QTL) mapping. Comparing the marker loci linked with the resistance to that of a mapped gene, predictions about gene identity can be made. The task of mapping unknown as well as novel sources of resistance did not begin until after chromosome maps were published in the 1990s. Several gene mapping studies have been conducted to identify QTL associated with stem rust resistance in wheat, including those that confer resistance to the Ug99 race group. The majority of these studies have aimed at mapping both all-stage resistance and APR that provide high level of resistance against the disease. Using different mapping population types, QTL mapping is usually carried out by testing the populations in different environments to get exposure of the wide diversity of the pathogen population. These mapping studies characterize the resistance, map the gene location, and identify markers that can be used in marker assisted selection for resistance breeding. The markers identified in a QTL mapping study are not always tightly linked with the gene of interest, and therefore require fine mapping. However, it must be noted that the availability and usability of these markers have been, and will continue to be a vital resources in resistance breeding against stem rust of wheat worldwide. Development

14

of tightly linked markers to these QTLs as well as characterization of the discovered QTL will play important roles in fighting the spread of rust in the coming decades. Gene Mapping Tools & Strategies Of different strategies implemented in mapping, early-generation (F2, F2:3) mapping (examples: Hiebert et al. 2011; Liu et al. 2011), mapping in recombinant inbred line (RIL) populations (examples: Bansal et al. 2008; Singh et al. 2013d), and in association mapping panels (examples: Yu et al. 2012; Letta et al. 2014) populate the literature. With regard to stem rust QTL mapping, biparental RIL populations seem to be preferred, because of 1) more power, and 2) the possibility of detecting rare alleles. The drawback of this approach, however, is that it takes time to develop the population, and unless the population size is adequately large, mapping resolution is typically low. A relatively new approach for mapping genes and QTL in several populations connected by a common parent has been developed in maize (Yu et al. 2008; McMullen et al. 2009). Coined as the nested mapping association (NAM) approach, this method scans for putative QTL in all biparental RIL populations with the genome of the common parent serving as a uniform background. The NAM approach combines the mapping power of a RIL population and higher mapping resolution of an association mapping panel, and therefore is able to discover both major and minor-effect QTL in a large population. Coupled with high throughput genotyping approaches, the NAM design has the potential to dissect marker-trait associations in several populations across several environments. The advent of better marker technologies such as randomly amplified polymorphic DNA (RAPD), amplified fragment length polymorphism (AFLP) and 15

simple sequence repeat (SSR) markers in the late 1990s paved the way for efficient mapping of several Sr genes. As a result, several stem rust resistance genes have been linked to diagnostic markers since then such as Sr39 (Gold et al. 1999), Sr24 and Sr26 (Mago et al. 2005), Sr9a (Tsilo et al. 2007), and Sr38 (Seah et al. 2000; Sharp et al. 2001). Diagnostic markers for several other important stem rust genes have been published (MAS Wheat 2014, http://maswheat.ucdavis.edu/protocols/), aiding to marker assisted gene pyramiding for resistance breeding. As the marker technology has evolved, SNP-based mapping approaches will likely dominate the former technologies in the future. Because of its improved cost/data point value, mapping genes using SNPs obtained from restriction-digestion based sequencing procedures have been demonstrated well in barley (Liu et al. 2014), and wheat (Saintenac et al. 2013a). Also known as genotyping by sequencing (GBS) or restriction site associated DNA (RAD) sequencing, this approach of obtaining population specific SNPs at a cheaper, faster rate can be used to enrich QTL regions linked to traits of interest (Chapters 2, 3 of this thesis). The flexibility offered by the GBS approach expands to imputation of missing haplotypes in related individuals due to low coverage in sequencing. Recent studies in several crop species including wheat have shown that imputing markers to construct missing haplotypes can be done with high accuracy (Rutkoski et al. 2013; Fu 2014). Thus, in mapping populations, where the individuals are either linked in families (F2, RIL populations) or via population structure (association mapping panels), the utilization of GBS SNPs appears promising.

16

Chapter 2

Mapping Putatively Novel QTL Conferring Adult Plant Resistance to Ug99 in the Biparental Population RB07/MN06113-8

The emergence and spread of the Ug99 race group of the stem rust pathogen in the past decade has exposed the vulnerability of wheat to this disease. Discovery of novel and effective sources of resistance is vital to reduce losses. The experimental breeding line MN06113-8 developed by the University of Minnesota wheat breeding program exhibits a high level of adult plant resistance (APR) to Ug99 race group. The University of Minnesota wheat cultivar RB07 exhibits a medium level of APR to Ug99 race group. Both of these lines are susceptible at seedling stages to the Ug99 races. To dissect the genetic mechanism of resistance present in these lines, MN06113-8 was crossed to RB07 to generate 141 F6 recombinant inbred lines (RILs). The RIL population was evaluated for APR to Ug99 in Kenya and Ethiopia over three seasons; and for APR to North American stem rust races in St. Paul, MN in one season. Composite interval mapping detected six quantitative trait loci (QTL) involved in APR to African stem rust pathogen races and three QTL involved in APR to North American stem rust pathogen races. One QTL located on the short arm of chromosome 2B was observed in all environments and provided resistance to all stem rust pathogen races, and could be a novel source of APR to stem rust pathogen races, including the Ug99 race group. Development of diagnostic

17

markers linked to this gene will facilitate will assist in marker assisted selection of resistant lines to generate varieties with elevated levels of APR to Ug99.

18

INTRODUCTION

One of the primary objectives of resistance breeding is the effective employment of resistance sources that promise durability. Constantly evolving pathogen populations challenge the effectiveness and durability of deployed resistance genes. The disease stem rust of wheat, caused by the fungal pathogen Puccinia graminis Pers. f. sp. tritici, is one such example where the pathogen is persistently evolving to overcome host resistance. One of the oldest plant diseases known to mankind (Kislev 1982), stem rust of wheat is highly destructive, and bears potential to completely destroy small scale farm plots to millions of hectares of susceptible varieties (Roelfs 1985b). The severe threat that stem rust has historically posed to global wheat production has been magnified in recent decades by the evolution of a highly virulent race, TTKSK. The isolate of this pathogen race was first observed in Uganda in 1998 and was named Ug99 for the country of origin and the year it was evaluated for its virulence phenotyping (Pretorius et al. 2000). Ug99 was found to overcome lines with Sr31, a widely deployed gene that provided resistance against stem rust of wheat at the time. Within a few years of its discovery, this race spread toward North Africa, West Africa, and the Middle East, and has potential to travel to West-South Asia as global wind patterns may transport the fungal spores over long distance (Singh et al. 2008a; Hodson et al. 2011). Ug99, later named as TTKSK after characterization using North American stem rust differential sets (Jin et al. 2008), along with its six other related races are virulent to 85–95% of breeding materials worldwide (Wanyera et al. 2006; Singh et al. 2011). The evolution of race 19

TTKSK and related races meant that the pathogen is capable of defeating multiple important rust genes that had been protecting wheat from stem rust for a long time, as evident by the breakdown of genes such as Sr31, Sr24, and Sr36 (Jin et al. 2008; Singh et al. 2008a; Jin et al. 2009a). Furthermore, many of the effective rust genes in bread wheat and its wild relatives were either ineffective against these races or unusable in the field with regards to grain quality and/or field performance (Singh et al. 2011). While this unexpected evolution of the Ug99 race group exposed the high vulnerability of wheat crops grown worldwide to stem rust, it also drew attention to the fact that the efforts in global rust monitoring and resistance breeding were not adequate to contain stem rust, making a search for durable resistance even more urgent. Durability of resistance in wheat is considered to be contributed by slow-rusting genes that protect the host based on the adult plant resistance (APR) mechanism. APR is assumed to act in ways that are nonspecific to rust races, and is generally distinguished by low infection frequency, reduced size of urediniospores, and overall diminished urediniospore production (Stuthman et al. 2007). In the case of wheat stem rust, APR has been further described as detected in mature plants, and mostly associated with absence of hypersensitive response (Hare and McIntosh 1979), and is quantitatively inherited (Knott 1982). APR genes are known to perform marginally when deployed individually. Singh et al. (2005) discuss that pyramiding a few APR genes can confer near immunity against diseases but may be difficult to accomplish because of the difficulties in phenotyping and identifying diagnostic markers associated with the resistance loci (Singh 2012). Combining multiple race-specific or major genes with or without APR genes has 20

also been proposed and utilized to obtain durable resistance against the disease (Ayliffe et al. 2008; Mago et al. 2011b; Evanega et al. 2014). Only a few stem rust APR genes have been discovered in wheat: Sr2 (Knott 1968), Sr55 (Lr67/Yr46/Pm46) (Herrera-Foessel et al. 2014), Sr56 (Bansal et al. 2014), Sr57 (Lr34/Yr18/Pm38) (Lagudah et al. 2006), and Sr58 (Lr46/Yr29/Pm39) (Singh et al. 2013c). Continual discovery of new sources that confer APR and promise durable resistance is vital for crop protection. The threat from Ug99 and its lineal races has been responded to, at least partly, by wheat research teams throughout the world by screening for resistance, and using the identified genes in their breeding programs. Many of these genes effective to Ug99 have been identified in non-bread wheat species. Examples include Sr32 which was identified in Aegilops speltoides (McIntosh et al. 1995), Sr37 in Triticum timopheevi (McIntosh and Gyarfas 1971), Sr39 in Aegilops speltoides (Kerber and Dyck 1990), Sr40 in Triticum araritum (Dyck 1992), Sr44 in Thinopyrum intermedium, and Sr53 in Aegilops geniculata (Liu et al. 2011). While wild relatives of bread wheat are excellent sources of previously undiscovered resistance genes, the issue of linkage drag, which occurs from introgression of genes derived from non-elite germplasm to elite breeding material, is a challenge. Breeders are usually hesitant to utilize such wild sources of resistance in their materials because of the time and effort it takes to select for lines with desired agronomic traits. Hence, discovery of resistance material in existing breeding programs would be a clear advantage, as crossing advanced lines with novel resistance to other elite lines would introduce the resistance while also preserving desired agronomic qualities.

21

The wheat breeding program at the University of Minnesota develops hard-red spring wheat varieties with superior agronomic performance and disease resistance. To safeguard the released varieties against a potential Ug99 epidemic, the program has routinely screened dozens of advanced experimental lines each year for resistance since 2005 in a stem rust nursery coordinated by USDA-ARS and the Kenya Agricultural Research Institute (KARI) in Njoro, Kenya. The advanced experimental line MN061138, despite being susceptible to Ug99 races at the seedling stage, was found to exhibit APR to Ug99 races the in Njoro stem rust nursery in Kenya. RB07 is moderately resistant – moderately susceptible (MRMS) to Ug99 races in the field, susceptible to Ug99 at the seedling stage. In order to understand the genetic mechanism of stem rust resistance in these lines, they were crossed (RB07/MN06113-8) to generate a biparental RIL population segregating for APR to Ug99 races. As both parent lines are highly inbred advanced lines, they possess several desirable agronomic traits. For the same reason, the resistance genes discovered in this RIL population can be used to develop varieties with elevated resistance to Ug99 races. In this study, we map the resistance segregating in the RB07/MN06113-8 RIL population. We also estimate the genetic effects of the detected loci and trace their origin. We hypothesize that some of the QTL regions detected in this study are putatively novel.

22

MATERIALS AND METHODS

Plant Materials A mapping population of 141 recombinant inbred lines (RILs) was developed via the single seed descent method by crossing ‘MN06113-8’ to ‘RB07’, both of hard red spring wheat growth habit. The line RB07 has the pedigree Norlander (PI 591623)/HJ98 and was developed by the University of Minnesota Agricultural Experiment Station. RB07 was released as a cultivar in 2007 on the basis of its high and consistent grain yield, earliness, resistance to wheat leaf rust, moderate resistance to Fusarium head blight, and good grain end-use quality (Anderson et al. 2009). The F6-derived line MN06113-8 has the pedigree MN97695-Lr52/HJ98-Fhb1 and is a breeding line at the University of Minnesota Wheat Breeding Program that advanced to second year yield trials before being discontinued for consideration as a new cultivar candidate due to low test weight and soft texture endosperm. Both parents are seedling susceptible to TTKSK, and TTKST races; and MN06113-8 is susceptible to race TTTSK of the Ug99 lineage (no data on RB07), and confer medium to high levels of resistance in the field in Kenya (Table 1). Seedling tests for reaction to the Ug99 races was carried out following the protocol outlined by Rouse and Jin (2011) at the United States Department of Agriculture, Agricultural Research Service (USDA-ARS) Cereal Disease Laboratory. Infection types (ITs) were recorded on a 0-4 scale according to Stakman et al. (1962).

23

Field Stem Rust Evaluation The F6:7 and F6:8 populations, along with the parents, were evaluated for their field response to African stem rust races at two locations in East-Africa over three seasons: at Njoro, Kenya during the main season from May - October of 2012 and the off-season from November 2012 to April 2013 (referred as Ken12 and Ken13 hereafter, respectively), and at Debre Zeit, Ethiopia during the off-season from March - June of 2013 (hereafter referred as Eth13). The population was also evaluated in St. Paul, MN, USA during May – August 2013 (hereafter referred as StP13) for response to N. American stem rust races. In the Njoro nursery, lines were planted in an augmented design with 1 check, ‘Red Bobs’. Each line was sown in double 70 cm long rows, 20 cm apart. On each side of the plot, and in the middle of the plots, a twin-row of susceptible spreader wheat cultivar ‘Cacuke’ was sown. The field was also surrounded by a border of several spreader rows comprised of susceptible wheat varieties which were artificially inoculated using a bulk inoculum of Pgt urediniospores collected at the Njoro field site; however wheat stem rust differential lines with known stem rust resistance genes indicated that the predominant, if not only, race present in the nursery since 2008 was race TTKST (avirulence/virulence formula on the wheat stem rust differential panel: Sr36, SrTmp/Sr5, Sr6, Sr7b, Sr8a, Sr9a, Sr9b, Sr9d, Sr9e, Sr9g, Sr10, Sr11, Sr17, Sr21, Sr24, Sr30, Sr31, Sr38, SrMcN) (Njau et al. 2010). In the Debre Zeit nursery, lines were planted in 1 m long twin rows that were flanked by spreader rows comprised of a mixture of susceptible wheat varieties 24

‘PBW343’, ‘Morocco’, and ‘Local Red’. Spreader rows were artificially inoculated with bulk inoculum of Pgt urediniospores to initiate the disease. At both locations, fresh urediniospores collected from susceptible varieties in the field were suspended in water followed by syringe-inoculation into susceptible spreader plants prior to booting stage (growth stages Z35 – Z37; Zadoks et al. 1974) at an approximate distance of 1 m. In the St. Paul nursery, lines were planted in hill-plots with 20 cm distance between the hills. The population was planted in an augmented design with 4 check varieties ‘Oklee’ (Anderson et al. 2005), ‘Thatcher’ (Hayes et al. 1936), ‘Tom’ (Anderson et al. 2012), and ‘Verde’ (Busch et al. 1996) planted after every 30 entries. A mixture of susceptible lines ‘Morocco’ and ‘LMPG-6’ were planted perpendicularly to surround the lines on all sides. To initiate disease, spreader rows were syringe-injected with a mixture of North American stem rust races MCCFC (isolate 59KS19), QFCSC (isolate 03ND76C), QTHJC (isolate 75ND717C), RCRSC (isolate 77ND82A), RKQQC (isolate 99KS76A), and TPMKC (isolate 74MN1409) at the jointing stage. The spreader rows were sprayed with a bulked mixture of Pgt races suspended in a light mineral oil suspension using an Ulva+ sprayer (Micron Sprayers Ltd., Bromyard, UK) after heading.

Phenotyping and Data Analysis Field reaction of the RILs to stem rust were recorded as disease severity on the 0to-100 modified Cobb scale (Peterson et al. 1948), and infection response, based on the size of pustules and amount of chlorosis and necrosis visible on the stem (Roelfs et al. 1992). Disease phenotyping of the population segregating for resistance was carried out after the susceptible check varieties in each trial had attained maximum severity. 25

Following Stubbs et al (1986), the severity response value was multiplied with the infection response to obtain coefficient of infection values. Using a mixed model (lme4 package in R 3.0.3, R Development Core Team, 2013), the growth stages of the lines were fitted to the coefficient of infection values to obtain fixed-effect estimates for each line, which were used for QTL mapping. Growth stages of each line in Kenya were determined mainly by assessing grain development stages such as watery, milky, soft dough, and hard dough; and also for stages of booting and flowering, as explained by Zadoks et al. (1974). Analysis of variance (ANOVA) was performed using function PROC GLM in SAS 9.1 (SAS Institute Inc., Cary, NC, USA) with genotype as a fixed effect, and environments and a combination of locations and years as random effects. Replicated checks were used to calculate the pooled error mean square value. Pearson correlation coefficients among the trials were calculated using the function PROC CORR in SAS 9.1.

Molecular Marker Assay Genomic DNA was extracted from ground seeds of the parents and F6:7 RILs using a modified cetyltrimethylammonium bromide (CTAB) protocol (Kidwell and Osborn 1992). The extracted DNA was quantified using an ND 1000 Spectrophotometer (NanoDrop Technologies, Delaware, USA). The population was genotyped using SNP markers obtained from two approaches: 1) the 9000 Infinium iSelect SNP assay (9K) (Cavanagh et al. 2013) and 2) genotyping by sequencing (GBS) (Elshire et al. 2011).

26

For genotyping using the Infinium iSelect assay, DNA suspended in ddH2O at approximately 80 ng/µl was submitted to the USDA-ARS Small Grain Genotyping Center, Fargo, ND, USA. The data generated was manually called using Illumina’s GenomeStudio 2011.1 (Illumina Inc., Hayward, CA). Briefly, monomorphic markers, markers with the same calls for the entire population, markers with more than 10% missing data, and markers that deviated from a 1:1 segregation ratio were discarded. Markers with 5% or less heterozygous calls were retained to avoid false purging of heterozygous loci. This resulted in 1,050 high quality markers that were retained for linkage mapping. To increase mapping resolution, and partly to investigate the feasibility of genome mapping using markers obtained from next-gen sequencing, the population was also genotyped using the GBS method (Elshire et al. 2011). In the GBS approach, a doubledigested library was created using the restriction enzymes PstI and MspI on 200 ng of DNA/sample, following Poland et al. (2012a) with modifications. Each library was 76plexed, with the parents repeated six times each, and the libraries were sequenced in two lanes of Illumina HiSeq 2000, generating 100 bp paired-end sequences. The sequences were processed using the UNEAK pipeline (Singh et al. 2013b) using the parameters -c 10 –e 0.025 to obtain de novo SNPs. Reads containing SNPs were used as query sequences and blastn-searched against the wheat chromosome survey sequences (CSS) to assign SNPs to unique chromosomes. The wheat CSS sequences are obtained by assembling reads obtained from sequencing flow-sorted wheat chromosomes from the ‘Chinese Spring’ variety (International Wheat Genome Sequencing Consortium, 27

http://wheaturgi.versailles.inra.fr/Seq-Repository/). To ensure that correct SNPs were obtained, only the full length alignment of a query sequence with the survey sequences allowing either one base mismatch or one gap was permitted. To circumvent retaining redundant SNPs on paralog sequences and duplicated regions among the A, B, and D subgenomes, SNPs thus obtained were filtered to remove those that mapped more than once to multiple chromosomes. SNPs that were monomorphic, had no allele calls for >10 individuals (>7% missing data), and were heterozygous in >10 individuals (7% heterozygosity) were also discarded. This process resulted in 932 high quality SNP loci that were retained for linkage mapping.

Linkage Map Construction and QTL Mapping SNPs obtained from both genotyping approaches (9K, de novo) were combined to assign markers to linkage groups. Linkage groups were constructed using Mapdisto V1.7.7.0.1 (Lorieux 2012) using a minimum logarithm of odds (LOD) value of 3.0. Genetic distances between the markers were calculated based on the Kosambi mapping function (Kosambi 1943). The program Windows QTL Cartographer 2.5_011, which implements composite interval mapping (CIM) to identify QTL, was used to analyze marker-trait associations (Wang et al. 2012b). The LOD threshold for declaring a significant QTL was calculated by 1,000 permutations at α=0.05. A walk speed of 1 cM was used for QTL detection. QTL effects were estimated as the proportion of phenotypic variance (R2) explained by the QTL. If multiple QTL were detected in an environment, digenic additive x additive epistatic interactions were tested among the detected QTL using the multiple interval mapping (MIM) algorithm available in the same program. 28

Using multiple marker intervals simultaneously, the MIM procedure fits multiple putative QTL directly in the QTL mapping model, and estimates several genetic architecture parameters including the effects of and interactions among significant QTL. Epistatic interaction among all SNPs, irrespective of their association with the detected QTL, was also carried out using the MIM algorithm. A QTL x QTL relationship was declared significant if LOD threshold was ≥ 1.0.

29

RESULTS

Disease Evaluation The disease pressure observed in all environments was adequate for good discrimination among stem rust phenotypes of the RILs, and the mapping of loci associated with quantitative resistance. The disease severity distributions skewed towards the lower percent severity responses overall (Figure 1). The highest disease pressure was observed in Ethiopia with lines showing up to 70% severity. Disease scores recorded for the RIL population along with the parents in all four environments are presented in Table 1. There was no significant difference between MN06113-8 and RB07 for their average stem rust responses across all environments (t-test P-value of 0.25 at α=0.05). This is not completely unexpected, as both parent lines exhibit similar levels of APR in the field (Table 1, Figure 1). Disease severity distributions for the RILs across all environments were continuous, suggestive of quantitative and polygenic resistance. Disease reactions of RILs between Kenya and Ethiopia nurseries were strongly correlated whereas correlations with the St. Paul reactions were lower, yet still significant (Table 2). The lower correlation coefficient between these environments suggests the presence of high genotype by environment (GxE) interaction, corroborated by the significant F-test value of 3.1 (significant at p-value 10 individuals (>7% missing data), or were heterozygous in >10 individuals (7% heterozygosity) were also discarded. SNPs obtained after these steps were converted to diploid format (AA/AB/BB) from allelic phases (A/C/G/T). This process resulted in 932 high quality SNP loci that were retained for linkage mapping.

Linkage Map Construction and QTL Mapping SNPs obtained from both genotyping approaches (9K, GBS) were combined to assign markers to linkage groups. Linkage groups were constructed using Mapdisto version 1.7.7.0.1 (Lorieux 2012) using a minimum logarithm of odds (LOD) value of 3.0. Genetic distances between the markers were calculated based on the Kosambi mapping 56

function (Kosambi 1943). Phenotypic data (stem rust severity) were collected on the RB07/MN06113-8 recombinant inbred line (RIL) population (Chapter 2 of this thesis). The program Windows QTL Cartographer 2.5_011, which implements the composite interval mapping (CIM) method to identify QTL, was used to analyze marker-trait associations (Wang et al. 2012b). A walk speed of 1 cM was used for QTL detection on linkage groups, and a QTL was declared to be present if the LOD threshold was calculated by 1,000 permutations at α = 0.05. QTL effects were estimated as the proportion of phenotypic variance (R2) explained by the QTL.

Imputation of GBS SNPs One problematic property of the GBS approach is the generation of substantial amount of missing genotype data (Poland et al. 2012a; Fu et al. 2014). Construction of linkage maps for genome mapping can be a difficult task with a dataset that is missing a significant portion of allele calls. In this dataset, the samples were 76-plexed, and therefore the issue of missing data was not egregious (< 7% missing data; see Chapter 2 of this thesis). However, to simulate scenarios where missing data could be a problem, the genotype matrix comprising of 932 GBS SNPs for the RIL population was modified to introduce missing allele calls. Missing values (‘NA’) were introduced randomly in the GBS dataset using R 3.0.2 (R Development Core Team, 2013) to simulate the genotype matrix with 20%, 30%, 40%, 50%, 60%, 75%, and 90% missing data. These datasets are hereafter referred as GBS20, GBS30, GBS40, GBS50, GBS60, GBS75, and GBS90, respectively.

57

Imputation of missing SNPs on the simulated data was done using principal component analysis (PCA) based imputation using the probabilistic PCA (ppca) algorithm in the R package ‘pcaMethods’ (Stacklies et al. 2007). The ppca algorithm first assigns row average values to the missing values, and then uses the singular value decomposition of the SNP matrix to create orthogonal principal components. In turn, the principal component values corresponding to the largest eigenvalues are used to reconstruct the missing SNP genotypes in the genotype matrix. The algorithm ‘ppca’ was chosen for its high imputation accuracy and efficiency in regards to the use of computational resources compared to other imputation algorithms of similar caliber (Moser et al. 2009; Fu 2014). In the algorithm, 25 PCA values were used to reconstruct all genotype matrices with missing data. Prediction values < 2 were assigned the homozygous genotype ‘1’ to represent alleles originating from the first parent; values equal to 2 were assigned the heterozygous genotype ‘2’; and values greater than 2 were assigned the homozygous genotype ‘3’ to represent alleles originating from the second parent. These values were assigned after careful and replicated manual scans of the imputed data to predict the true genotypes as much as possible.

Summary Statistics of Genotype Matrices Both non-imputed and imputed datasets were analyzed using PowerMarker V3.25 (Liu and Muse 2005) to estimate population statistics. The method of moments estimator was used to estimate the within-population inbreeding coefficient and relatedness between the individuals (Ritland 1996). An unbiased estimator that uses the inbreeding coefficient calculated using the method of moments in the previous step was used to 58

calculate gene diversity. Polymorphism information content (PIC), a diversity measure among the individuals in a population, was calculated according to Botstein et al. (1980). Estimation of these population statistics was done with 10,000 nonparametric bootstraps across different loci at a confidence interval of 95% (α = 0.05). The number of recombinations in each RIL on all datasets (9K, GBS-unimputed, and GBS-imputed) was estimated using the R package ‘hsphase’ (Ferdosi et al. 2014). We were also interested in knowing the distribution of recombination sites across the genome, and specifically the distribution within chromosomal arms. Although recombination in the wheat genome differs by chromosome, gene-rich regions (GRRs) are known to be the recombination hotspots (Sandhu and Gill 2002; Sidhu and Gill 2004). In general, studies report that most recombinations occur on distal 20-50% of both long and short arms of the chromosomes (Lukaszewski and Curtis 1993; Faris et al. 2000). Therefore, recombination events were calculated on the distal 40% of each arm versus the remaining region on each chromosome. The distal 40% of each arm was determined as 40% length of the genetic distance of the chromosomal arm.

Methodology and Workflow Comparison The cost, time, and resources required to genotype the RIL population using both methods were compared so that some characteristics between the two methods could be understood to determine their usability in programs and/or projects that are similar to ours. The discussed cost estimate strictly pertains to the genotyping cost, and does not include the cost of the manual labor involved. The amount of time required to genotype the population is the ‘active’ time used in preparation of the DNA samples, sequencing, 59

and data analysis; and does not include the latent time between procedures. Computational resources and skills needed to analyze the data obtained from both procedures are also briefly discussed.

60

RESULTS AND DISCUSSION Genotype Properties The genotype calls obtained from both the 9K SNPs and the GBS SNPs were first analyzed for deviation from an expected segregation ratio of 1:1. However, as described in the ‘Materials and Methods’ section, the 9K dataset (1,050 SNPs) was devoid of any markers deviating from the 1:1 segregation ratio as SNPs that deviated from the ratio were discarded. Of the 932 GBS SNPs, 164 SNPs were found to be skewed towards either parental genotype. Of these 164 SNPs, 48 SNPs were over-representative of the MN06113-8 genotype, whereas 116 SNPs were over-representative of the RB07 genotype. Overall, 49.5% of marker genotypes were inherited from MN06113-8 in the 9K data, with 50% inherited from RB07, and 0.5% heterogyzous genotypes. In the GBS data, 46% of the marker genotypes originated from MN06113-8, 49% from RB07, 3% were heterozygotes, and 2% had missing data. The RILs used for genotyping were inbred to F6:7 generation, and as such, only 1.6% of genotype calls in the population were expected to be heterozygous; and these numbers observed in both genotyping methods are expected of a highly inbred population. These results also indicated a higher proportion of heterozygous and missing allele calls in the GBS dataset, as permitted during the filtering of genotype calls.

Linkage Groups Construction Linkage groups for both 9K and GBS SNPs datasets were constructed using the same parameters in Mapdisto version 1.7.7.0.1 (Lorieux 2012). Of 1,050 9K SNPs, 964 SNPs were placed in 30 linkage groups representing 18 wheat chromosomes. 61

Chromosomes 2D, 4D, and 7D were not represented by any SNPs. The number of SNPs per linkage group ranged from 2 (chromosomes 2A, 3D, 6B) to 145 (chromosome 5B) with an average of 32 SNPs. The 964 SNPs distributed over the 30 linkage groups covered a total of 1,294 cM of the wheat genome. The sizes of the smallest and the largest linkage groups were 0.4 cM and 168 cM, respectively, with an average size of 43 cM. Similarly, 925 of the 932 GBS SNPs were assigned to 31 linkage groups that represented all 21 wheat chromosomes. The number of SNPs in these linkage groups ranged from 2 (chromosomes 4A, 6B, 6D, 7D) to 131 (chromosome 2B) with an average of 30 SNPs per linkage group. The smallest linkage group was 0.8 cM, the largest group was 147 cM long, and the average size of all linkage groups was 42 cM. The size of the wheat genome covered by the 31 linkage groups constructed using 925 GBS SNPs was 1,305 cM. Properties of the linkage groups constructed using the two genotyping methods are presented in Table 1. One particular advantage GBS has over the SNP chip based genotyping is the slightly better coverage of the wheat D genome. The wheat D genome is often the least represented in genotyping platforms, owing to its lower frequency of polymorphic sequences (Chao et al. 2009; Allen et al. 2011). While the number of markers mapped to the D genome in our GBS dataset is not large compared to the A and B genomes, the retrieval of all seven D chromosomes during linkage mapping is a good indication that the GBS approach can be manipulated to obtain more SNP markers from the D genome. One possible way to achieve this is by using less stringent filtering parameters than the ones used in our study. However, a more reliable approach may be to map the reads 62

obtained from sequencing of GBS libraries to the respective sub-genomes of wheat in order to obtain sub-genome specific polymorphic markers. With two of three wheat’s subgenomes already sequenced the A genome from wheat’s diploid ancestor Triticum urartu (Ling et al. 2013) and the D genome from Aegilops tauschii (Jia et al. 2013) – subgenome specific mapping of sequence reads would help to retain reads that would otherwise be discarded during SNP-calling due to mismatches among the reads originating from the A, B, and D sub-genomes.

Recombination and Genome Coverage To better understand the distribution of parental genotypes in the population, the number of recombination events in each RIL was estimated. The average number of recombinations per RIL in both 9K and GBS genotype matrices was 27. Seventy-eight (55.3%) individuals had more recombinations than the average in GBS dataset, compared to 65 (46.1%) in the 9K dataset. Only 2 individuals had the same number of recombinations in both matrices. The total number of recombinations in the 9K matrix was 3,746 which was slightly lower than that in the GBS matrix, at 3,790. To visualize these results, recombination blocks per line observed in both genotype matrices were plotted against the SNPs in each chromosome (Figure 1). As seen in the figure, both congruous and incongruous patterns of recombination blocks exist between the two genotype matrices. The difference in reported recombination breakpoints likely arises from the difference in assay design between the two genotyping methods as different parts of the genome might have been sampled, which alter the genome coverage in these two genotyping methods. This can potentially lead to over- or 63

under-representation of different haplotype matrices, which results in the detection of different recombination sites and number of recombination events. The difference in genome coverage is also corroborated by the difference in properties of the linkage groups constructed using SNPs obtained from the two methods. This is illustrated in Figure 1C where the portions of the genome represented by the two methods are compared. The figure shows the differences in genome sampling between the two genotyping methods in one RIL (‘MN06_01’) across the chromosomes, especially visible among the chromosomes 1A, 1B, and 5A. To visualize distribution of 9K SNPs and GBS SNPs along the linkage maps, linkage maps constructed separately using 9K and GBS datasets were compared with the linkage map constructed using a combined ‘9K+GBS’ SNPs (see Chapter 2 for more details). The results support the patterns of haplotypic distribution as observed in Figure 1C. The 9K SNPs are located away from the GBS SNPs on chromosomes 1A, 1B, and 5A in combined linkage map, owing to the differences in genome sampling between the two SNP types (Supplemental File 1). On chromosomes that show similar sampling of the genomic regions (such as 2B, 3A, and 4B), both 9K and GBS SNPs are intercalated in proximity to each other in the combined ‘9K+GBS’ linkage maps (Supplemental File 1). This difference in SNP distribution between the two datasets illustrates that dissimilar portions of the genome are represented by the two genotyping approaches. Such differences are also observed in all RILs in the population in general (data not shown). We believe that the accumulation of such differences over the whole genome across the population is the reason behind the observation of incongruous recombination patterns. 64

As expected, the distribution of recombination events favored the distal ends of the chromosomes compared to the centromeric regions in both datasets. In the GBS dataset, 1,523 recombination events were observed on the distal 40% of both arms (623 on short arms, 900 on long arms), and 1,257 in the centromeric region. Similarly, 843 recombination events were observed on the distal 40% of both arms in the 9K dataset (396 on short arms, and 447 on long arms), with the centromeric region recording 706 recombinations. In our population, we observe that the distal ends have 10% more recombinations in the GBS dataset, and 9% more recombinations in the 9K dataset, relative to the centromeric regions in each respective dataset. The highest number of recombinations among all three sub-genomes in the GBS dataset was observed in the Bgenome (1,608 recombinations) whereas the A-genome recorded the most recombinations in the 9K dataset (1,791 recombinations). The D-genome in both datasets had the fewest recombination events with 691 recombinations in the GBS dataset, and 369 in the 9K dataset. To uncover the genetic architecture controlling the traits of interest, the use of molecular markers representative of the genome is important in gene mapping studies. Markers that are significantly linked to the trait can provide remarkable improvements in breeding for allele enrichment and trait improvement. Since the choice of genotyping platform can impact the quantity of molecular markers and their distribution in different genomic regions, understanding the differences in marker-related genome properties can assist in such choice among the different available genotyping approaches. In our investigation of such properties between the two genotyping methods discussed here, the 65

difference in number of polymorphic markers identified and used in creating linkage groups was not strikingly different; whereas the distribution of those markers in different genomic regions as shown by the difference in sites of recombination was noteworthy. The difference in genome coverage is also corroborated by the difference in properties of the linkage groups constructed using SNPs obtained from the two methods, and the differences in QTL detection between the two methods, as described in upcoming sections.

QTL Mapping Mapping of QTL associated with resistance to stem rust of wheat was carried out in the RIL population using stem rust severity data collected at three locations over four seasons. This was done to assess the impact on QTL mapping using linkage groups constructed from markers obtained from the two genotyping approaches. For detailed information on disease phenotyping and data statistics, see Materials and Methods, Chapter 2 of this thesis. The CIM method of QTL mapping detected nine QTL in the 9K dataset on linkage groups representing the chromosomes 2B, 3A, 4A, 4B, 5B, and 6D (Appendix I). Similarly, eight QTL distributed on chromosomes 1A, 2A, 2B, 2D, 4A, 4B, and 7A were detected in the GBS dataset (Appendix I). Four QTL detected in the 9K dataset, namely the QTL on chromosomes 2B (QSr.umn-2B.1, QSr.umn-2B.2), 4A (QSr.umn-4A), and 4B (QSr.umn-4B.2) were also detected in the GBS dataset. The similarity of the QTL was determined based on their location, parent contributing the QTL allele, and SNPs associated with the QTL (Table 3, Chapter 3 of this thesis). QTL not common between 66

the 9K and GBS datasets suggests that the portions of the genome represented by the SNPs obtained from the two genotyping approaches are different, as we discussed earlier. However, uniform genome coverage, inclusion of more environments, or highly correlative phenotype data between the environments might absolve this issue. On the whole, both methods performed reasonably well, and were able to predict the location and effect of QTL involved in stem rust resistance.

Imputation of GBS Markers One advantage of the GBS approach over the chip-based genotyping approach is the flexibility of the system that allows some control over the number of polymorphic markers that can be generated among lines in a population. An imputation study has not been conducted in an RIL population yet, therefore we studied the effect and accuracy of genotype imputation on the GBS dataset after simulating datasets with several successive missing proportions. The highest imputation accuracy was observed when the proportion of missing data was the lowest (Figure 2). The missing genotypes were predicted with accuracy of 96% when the dataset was missing 20% of the genotypes. The accuracy of genotype imputation was reduced as datasets had a higher proportion of missing data, with a major drop-off in datasets > 60% missing data, which had imputation accuracies of < 84%. The effect of missing data did not appear to impact population parameters such as gene diversity, polymorphism information content (PIC), and the amount of heterozygosity. An overall decreasing trend for each of these population characteristics can, however, be observed with increasing missing values in the dataset. The inbreeding coefficient had an 67

overall increasing trend as datasets had more missing proportion of genotype calls. There was no significant change in the allele type from imputation in the imputed datasets, except in the GBS90 dataset where the proportion of RB07 alleles increased by 3%. Our dataset contained only 932 high-quality GBS markers, which is sufficient for gene mapping studies, given the high LD of hexaploid bread wheat (Chao et al. 2007; Chao et al. 2009). However, the marker imputation results presented here show that more de novo markers obtained from the GBS approach with moderate levels of missing allele calls (up to 40%) may be used in order to obtain higher resolution in genome mapping studies. With as much as 40% missing data, the genotype matrix can be predicted with imputation accuracy of 90% or higher. While this may potentially introduce some biases in the study, the level of gene diversity, PIC, and heterozygosity were not significantly altered, implying that the population does not deviate significantly from the expected levels of inbreeding. This is illustrated in Figure 2, where the inbreeding coefficient increases negligibly from non-imputed GBS dataset to GBS90 dataset. The observed slight increase is most likely due to the introduction of false genotype calls in imputed datasets with a higher proportion of missing data. In most crop breeding programs, higher level of inbreeding is desired so that the allelic composition of genes that control the desired traits in the breeding germplasm may be preserved. Yet, the consistency in inbreeding levels among the imputed datasets observed here should not be the only determinant behind the use of a dataset with large amounts of missing allele calls. Datasets with large amounts of missing data introduce severe problems such as inaccurate and inflated linkage groups, and erroneous QTL detection. 68

QTL Mapping Using Imputed Datasets The effect of marker imputation in construction of linkage groups and QTL mapping in a biparental population has not been investigated yet. Thus, we used imputed GBS datasets, as described in the Materials and Methods section, to investigate QTL mapping. Construction of linkage groups and QTL mapping, however, were carried out using the imputed dataset with 40% and 75% missing data only, which had imputation accuracies of 90% and 71%, respectively. Twenty-nine linkage groups were formed in the GBS40 dataset, and covered 4,785 cM of the genome (Table 2). Similarly, 30 linkage groups were formed in the GBS75 dataset, covering 14,879 cM of the genome. Genome sizes represented by these linkage groups are approximately 3.5 times and 11 times larger than the size covered by the original, non-imputed GBS dataset. This increase in genome size is most likely due to the introduction of inaccurate genotype calls in the dataset with the higher proportion of missing values. Inflation of linkage groups by introduction of false genotype calls is also supported by the average marker interval distance of 5 cM in the GBS40 dataset and 19 cM in the GBS75 dataset. Therefore, marker imputation in datasets with a large proportion of missing data appears to introduce errors, and as such, avoiding the use of such datasets is pragmatic. As the imputation accuracy dropped, more markers were also unlinked to any linkage group (Table 2). Two linkage groups (chromosomes 3D and 7D) were not detected in the GBS75 dataset. The number of QTL discovered in both GBS40 and GBS75 datasets using the CIM approach was 8 and 7, respectively (Table 2). However, most of the QTL detected 69

in these datasets were different relative to the non-imputed dataset, with only one consistent QTL among all datasets (Appendix I). The large-effect QTL observed on chromosome 2B in the non-imputed GBS datasets was the only consistent QTL in the GBS40 and GBS75 datasets, with lower accuracy of the QTL positions in the imputed datasets (Figure 3, A-C). Only the GBS40 dataset correctly predicted the same SNPs (TP24441 and TP17690) linked to the large-effect QTL as predicted in the non-imputed GBS dataset. Although the large-effect QTL is still detected in the GBS75 dataset, the QTL was incorrectly placed near the end of the linkage map; and the significant markers TP24441 and TP17690 were incorrectly assigned wrong map positions. Both imputed datasets predicted the percentage of phenotypic variation and allelic effect similarly to the non-imputed dataset. The small-effect QTL observed in the Kenya 2012 environment was detected in both imputed datasets, although its position and linked markers in GBS40 and GBS75 were different relative to the non-imputed dataset. It is likely, given the inflation in size of linkage groups experienced with imputing, that most of the QTL detected in the imputed datasets are inaccurate. While validation of the detected QTL would provide a definitive answer, the results indicate that large-effect QTL can be detected if the dataset comprises a large proportion of imputed genotypes. The small-effect QTL may also be detected, but such prediction might not necessarily be accurate.

Comparison of Methodology and Required Resources With regard to the methodologies and workflow, chip based genotyping is relatively easier than the GBS approach, though the latter had a faster turnaround time in our case. The 9K genotype calls were obtained from the USDA Genotyping Facility at 70

Fargo, ND that needed manual inspection using Illumina’s GenomeStudio program version 2011.1 (Illumina Inc., Hayward, CA) before use. In the GBS method, DNA libraries for sequencing were generated in-house in less than two days. The sequences were filtered based on barcodes and trimmed before calling SNPs in the population. In addition, sequence alignment to the WCSS and further data parsing was required. Thus, the need for high-end computational resources and bioinformatic expertise (ability to work in unix environment, as well as programming skills) is essential in the GBS approach to manage and work with the large amount of sequence data generated from parallel sequencing. On average, 256 gigabytes (gb) of memory was requested on any available node with the Minnesota Supercomputing Institute (MSI, https://www.msi.umn.edu/) while working with the GBS procedures such as quality control of the sequences, SNP calling, and sequence alignment. Similarly, several hundred gb of hard disk space was needed to store the sequence files and any output files created during the procedures mentioned above. The SNP chip based method, however, required less computational resources but a proprietary program (GenomeStudio) was needed to visualize and analyze the data generated from the genotyping assay. The program was run on a Windows machine with 64 gb of available memory. The hard disk space requirement to store the data files was less than 50 megabytes. The biggest advantage of the GBS approach however, is perhaps the economical aspect of this method. In our study, we obtained a comparable number of usable SNPs for QTL mapping from both genotyping approaches (964 from 9K, 925 from GBS). The cost per 9K SNP used for mapping was approximately $8.20, whereas the cost per GBS SNP 71

used in QTL mapping was approximately $2.10. These figures are exclusive of the labor cost, in which the GBS method is also advantageous over the chip-based genotyping method. Although several filters can be applied within the program GenomeStudio to parse the genotype calls obtained using the 9K chip, the genotypes still need to be manually inspected for each SNP between the two parents in order to re-cluster the individuals to distinct genotype groups. In our dataset, 2,524 polymorphic SNPs were obtained after applying the filter to remove SNPs monomorphic between the two parents. Each of these SNPs had to be inspected to assign the correct genotype calls to the RILs. At the inspection rate of approximately three SNPs every two minutes, the total time required to tag the population with correct genotype calls was approximately 28 hours. As the program did not allow correction of the incorrect genotype calls, the exported data had to be edited to assign final genotypes to each individual. On the other hand, the GBS procedure required less than 3 hours to obtain the final genotype calls. Creating the input file with each individual labeled with the barcode used during library preparation was essential before running the SNP-calling program UNEAK. This task was completed in about 30 minutes, followed by approximately one hour to obtain the SNPs from UNEAK. As the allele calls were reported in base format (AA/CC/GG/TT), they were converted to a biallelic format (AA/BB), which was accomplished in about 30 minutes. While the time needed to troubleshoot errors that appeared during both procedures has not been discussed here, the GBS approach was more efficient because of its lower economical burden and advantages in automated data processing. Yet, as we indicated in the results, the genome areas targeted by these approaches are slightly different, which may lead to 72

detection of some different QTL between the two methods. If marker coverage were better, and perhaps uniform, between the two methods, these differences should disappear. Based on the comparative analysis, we are confident that the GBS approach can be used as an independent, standalone approach for genotyping and conducting genome studies.

73

CONCLUSION At present day, the cost of genotyping is lower than ever. Highly economical genotyping approaches provide a user the options of using different types of highthroughput genotyping methods, ranging from simple sequence repeat based genotyping to sequence-based genotyping. In our study, we compared two high throughput genotyping methods used in genomic studies of wheat: SNP-chip assay, and genotyping by sequencing (GBS). The results showed that both methods are powerful means of studying the genome and provide enough resolution to carry out marker-trait association studies. The key attributes of high interest to a researcher might be the cost and data turnaround time, in which the GBS approach bests the SNP-genotyping method. The GBS approach was also able to provide a better coverage of the wheat genome, including that of the often poorly represented D-genome. The SNP-chip based genotyping however requires less computational knowledge and resources to process the data. The choice of the genotyping platform for gene mapping and other genome studies may come down to the question of cost and available resources.

74

Table 1: Results of linkage groups formation using the 9K and GBS SNPs. A. Method

9K

Chromosome Linkage groupsa SNPs

b

Size (cM)

c

1A

2A

3A

4A

5A

6A

7A

1B

2

4

1

1

3

2

1

2

76

32

138

26

23

121

43

75

38

156

61

31

76

105

2B

3B

4B

1

1

1

17

110

7

57

141

47

Total 5B

6B

7B

1D

2D

3D

4D

5D

6D

7D

1

4

1

1

0

1

0

1

2

0

82

145

12

66

23

0

2

0

21

18

0

94

168

34

95

73

0

1

0

24

18

0

30

SNPs per sub-genome

459

439

64

962

Size per sub-genome

542

636

116

1,294

B. Method

GBS

Chromosome

1A a

Linkage groups

2A

3A

4A

5A

6A

7A

1B

2B

3B

4B

Total 5B

6B

7B

1D

2D

3D

4D

5D

6D

7D

1

2

1

2

1

3

1

2

1

1

1

1

2

1

1

1

1

2

2

2

2

SNPsb

61

29

56

63

15

87

80

62

131

40

61

117

11

46

10

16

3

11

13

9

4

Size (cM)c

95

45

146

58

23

36

98

80

111

43

90

147

25

83

29

44

23

35

64

12

18

31

SNPs per sub-genome

391

468

66

925

Size per sub-genome

501

579

225

1,305

a

Number of linkage groups formed for each wheat chromosome

b

Number of SNPs that mapped to all linkage groups representing each chromosome

c

Size of the linkage groups combined if >2 linkage groups were observed for a chromosome 75

Table 2: Comparison of linkage mapping results among the 9K dataset, non-imputed GBS dataset, and datasets with 40% and 75% missing allele calls. Methoda 9K GBS GBS40 GBS75

SNPsb

Linkage Groups

1,050 932 932 932

30 31 29 30

SNPs (%) in Linkage Groups

Unlinked SNPs (%)

Genome Size (cM)

QTL

8.2 0.8 1.2 10.8

1,295 1,306 4,785 14,879

4 8 8 7

91.8 99.2 98.8 89.2

a

GBS40 = imputed GBS dataset with 40% missing allele calls; GBS75 = imputed GBS dataset with 75% missing allele calls

b

Number of polymorphic markers used in linkage mapping and imputation of missing genotype data

76

A.

B.

77

C.

78

Figure 1: Recombination blocks observed in 9K (Panel A) and GBS (Panel B) genotype datasets; and a representative example of the difference in genome sampling between the 9K and GBS methods (Panel C). In Panels A and B, the population is arranged in descending order on the Y-axis and SNPs are arranged by the chromosomes they belong to on the X-axis separated by the white vertical bar. Panel C represents an example of the difference in genome sampling between the 9K and GBS methods in the recombinant inbred line (RIL) ‘MN06_1’ across all 21 chromosomes. The size difference within each chromosome (for example, within 1A_9K and 1A_GBS) is due to the differences in number of SNP markers between the two methods that are distributed along the chromosome. No linkage groups were obtained for chromosomes 2D, 4D and 7D using the 9K SNPs. In all panels, the colors gray and black represent MN06113-8 and RB07 haplotype blocks, respectively whereas the white dots (white vertical lines on panel C) indicate missing data.

79

1.0

20000

0.9

18000

0.8

16000

0.7

14000

0.6

12000

0.5

10000

0.4

8000

0.3

6000

0.2

4000

0.1

2000

Genome Size (cM)

Frequency

Gene Diversity Heterozygosity PIC Inbreeding Coefficient Imputation Accuracy Major Allele Frequency Genome Size (cM)

0

0.0 0

20

30

40

50

60

75

90

Percent of missing alleles in the GBS dataset

Figure 2: Characteristics of genotype matrices in imputed and non-imputed datasets. ‘0’ represents the original GBS dataset with no missing allele calls introduced. The genome sizes (sizes of linkage groups summed together) of imputed and non-imputed GBS datasets are shown on the secondary Y-axis to the right.

80

Chromosome 2B

81

TP4593 TP7605 TP43482 TP38526 TP18814 TP40207 TP1839 TP23420 TP6436 TP38912

TP24441 TP17690

315 329.4 330.5 331.2 333.9 339.9 342.9 345.9 346.6 349

301.1 308.7

18 95.7

TP17690

TP4593 TP38912 TP6436 TP1839 TP23420 TP43482 TP38526 TP7605 TP40207 TP18814

101.8 105.5 105.9 106.3 107.1 107.5 107.9 108.3 109.1 109.4

94.5

90.9

TP48796

TP24441

TP46931 0 TP26772 1.5 TP46179 2.3 TP6864 3 TP5073 3.4 TP33150 3.8 TP28068 4.6 TP3836 8.7 TP5413 9.4 TP48783 11 TP44603 13.3 TP46799 14 TP7529 30.2 TP4248 32.2 TP586 34.6 TP29297 35.7 TP67 36.8 TP15279 37.2 TP14831 37.6 TP26488 38 TP37517 39.1 TP23986 40.3 TP35021 41.5 TP19149 42.3 TP46334 42.7 TP30211 43 TP22013 44.2 TP34112 45 TP48612 45.4 TP42224 45.8 TP18308 46.2 TP25744 46.9 TP36482 47.3 TP34283 47.7 TP47506 48.5 TP25235 72 TP5574 73.2 TP35941 73.6 TP49699 78.8 TP25302 79.1 TP27408 79.5 TP28314 79.9 TP43861 80.3 TP1719 88.9

LOD Values 16

TP44603 0 TP6864 8.5 TP46931 14.4 TP26772 31.7 TP28068 38.2 TP46179 41.8 TP5073 44.9 TP33150 48.9 TP5413 52 TP3836 54.6 TP48783 56.9 TP46799 64.2 TP7529 91.9 TP4248 101.6 TP26488 105.4 TP35021 109.6 TP46334 110.7 TP19149 111.5 TP34283 119 TP47506 121.4 TP36482 133.2 TP42224 137.7 TP48612 138.4 TP18308 139.9 TP25744 145.9 TP22013 150.1 TP34112 154.2 TP30211 155.3 TP37517 163.4 TP15279 169.1 TP586 187.3 TP29297 190.3 TP14831 194.5 TP67 195.6 TP25235 223.1 TP5574 226.2 TP35941 228.5 TP25302 254.3 TP49699 259.2 TP43861 263.3 TP28314 269.5 TP27408 275.8 TP1719 295.2

LOD Values

A.

GBS Ken12

14 Ken13

Eth13

12 StP13

10

8

6

4

2

0

Chromosome 2B

B. GBS40

16

14 Ken12 Ken13 Eth13 StP13

12

10

8

6

4

2

0

C. 12

Ken12

GBS75

Ken13 Eth13

10

StP13

6 4 2

1628.7 1655.4 1702.4 1724.5 1769.1 1791.1 1889.6 1948.7 1979.3 2005.7 2043.3 2064.1 2086.7 2109.8 2127.4 2141.2 2154.6 2189.5 2198.6 2231.3 2302.9 2342

TP24441

TP48796 TP46179 TP46931 TP4593 TP48783 TP33150 TP27408 TP28068 TP25302 TP5574 TP35941 TP28314 TP25235 TP46334 TP34283 TP36482 TP48612 TP14831 TP18308 TP67 TP26488 TP1719

TP23420 TP6436 TP43482 TP1839 TP43861 TP49699 TP7605

TP17690

1203.3 1265.8 1289.9 1378.8 1492.5 1504.2 1572.1

1177.2

1609.8

0 TP38912 22 TP40207 32.2 TP18814 52.2 TP38526 86.8 TP3836 123.3 TP7529 273.6 TP4248 300 TP44603 383.7 TP586 468.9 TP26772 529.1 TP47506 572.4 TP25744 637.2 TP42224 671.8 TP22013 687.2 TP35021 700.1 TP37517 711.7 TP19149 722 TP15279 744 TP29297 793.2 TP23986 822.3 TP30211 857.8 TP34112 867.4 TP46799 926 TP5413 962.1 TP5073 1066.7 TP6864 1119.9

LOD Values

8

Chromosome 2B

Figure 3: A comparison of the QTL detected on chromosome 2B among the three GBS datasets: non-imputed GBS dataset (Panel A), GBS40 (Panel B), and GBS75 (Panel C). The Y-axis indicates the logarithm of the odds (LOD) values with the dotted line representing the threshold LOD score of 2.5. The X-axis is labeled with SNP markers and the genetic distances in centimorgans (cM) between the markers. The significant SNP markers (TP24441 and TP17690) in non-imputed GBS dataset are labeled in red color (and black triangles) in all three panels to indicate their positions relative to the QTL peaks. Note the increase in sizes of the GBS40 and GBS75 linkage groups compared to that of non-imputed GBS dataset. In comparison with the non-imputed GBS dataset, the order of the SNP markers also changes in linkage groups constructed using genotype information in imputed datasets with higher proportion of missing alleles. 82

Chapter 4

Nested Association Mapping of Stem Rust Resistance in Wheat Using Genotyping by Sequencing

Nested association mapping (NAM) is an approach to map trait loci in which families within populations are interconnected by a common parent. By implementing joint-linkage association analysis, this approach is able to map causative loci with higher power and resolution compared to biparental linkage mapping. The recently developed genotyping by sequencing (GBS) approach is a relatively fast, efficient and cost-effective method of genotyping a large number of individuals. Additionally, the GBS method allows for discovery of de novo markers that are specific to the individuals or populations under study. We combined the NAM approach with GBS to dissect and understand the genetic architecture controlling stem rust resistance in wheat. Ten stem rust resistant wheat varieties were crossed to the susceptible line LMPG-6 to generate F6 recombinant inbred lines (RILs). The RIL populations were phenotyped at four environments in Kenya, South Africa, and St. Paul, Minnesota, USA. We identified several minor-effect QTL contributing towards adult plant resistance (APR) to North American Pgt races as well as the highly virulent Ug99 race group. Validation of markers that are significantly associated with each QTL is necessary to generate diagnostic markers for marker assisted resistance breeding. The usefulness of GBS-derived de novo SNPs in mapping APR to

83

stem rust shown in this study could be used as a model to conduct similar marker-trait association studies in other plant species.

84

INTRODUCTION Since the earliest days of wheat cultivation, the disease stem rust of wheat caused by the fungus Puccinia graminis f. sp. tritici (Pgt) has been a major constraint to wheat production in many areas of the world. Historical records show that the disease is highly damaging during epidemics and is capable of posing a serious threat to world food security. The pathogen is well known for its ability to travel long distances and capacity to evolve at the site of deposition (Hodson et al. 2011). Genetic resistance has been the main means of fighting this disease primarily because of its effectiveness. However, the pathogen has historically proven that it is able to overcome the deployed genes, giving rise to disease epidemics in regions with susceptible cultivars (Roelfs 1985a; McIntosh et al. 1995). The recent evolution and spread of the virulent African stem rust race TTKSK and its derivative races are the perfect example of this phenomenon. Race TTKSK and six other lineal races of the same group have already defeated resistance genes that had been effective for several decades (Mago et al. 2011a). As predicted by wind trajectories, the arrival of these stem rust races in the breadbaskets of the world is likely. Also, the Pgt races in North America are changing, as documented by detection of a highly virulent race, TTTTF, only a few years ago (Jin 2005). Therefore, discovery of novel sources of resistance and their deployment is an essential process in order to fight off this constantly evolving pathogen. The nature of genetic resistance to stem rust of wheat is mainly qualitative, in the form of major genes derived from hexaploid bread wheat and related species. All-stage resistance (also known as seedling resistance) has been a significant part of rust 85

resistance breeding. One of the first resistance genes to be deployed, Sr12 in the University of Minnesota cultivar ‘Thatcher’ protected the majority of the wheat acreage from rust epidemics in the northern US wheat growing regions in the 1930’s and 1940’s (McIntosh et al. 1995). Varieties with the gene Sr31, bred and distributed by CIMMYT from the mid 1960s, were popular globally until they were defeated by the highly virulent Pgt race TTKSK (aka isolate Ug99) in 1998 (Pretorius et al. 2000). TTKSK, along with six other races in the Ug99 lineage, has defeated almost the half of the discovered stem rust resistance genes (Jin et al. 2008; Jin et al. 2009b; Mago et al. 2011a; Pretorius et al. 2012), and the majority of deployed genes. This is one example among several cases where Pgt races have evolved to defeat several important stem rust resistance genes that had been protecting the wheat crop from this disease. As the selection pressure on the pathogen leads to its evolution and renders genes ineffective, the threat of crop loss from possible rust epidemics increases. In addition to seedling resistance, adult plant resistance (APR) is considered to be effective against a wider array of Pgt races, and are assumed to be durable, mainly because of their nonspecific effectiveness (Lindhout 2002; Stuthman et al. 2007). APR genes are expressed mainly during the adult plant stages, and provide APR during important plant growth and development phases. While QTL mapping studies have reported several QTL providing APR against Pgt races, only five APR genes (Sr2, Sr55, Sr56, Sr57, and Sr58) has been discovered to date. As APR genes contribute less additive effectiveness compared to seedling genes, the level of resistance conferred by these genes in the field is usually inadequate. Thus, pyramiding both APR and seedling genes to 86

obtain durable APR is essential for effective disease control (Singh et al. 2000; Evanega et al. 2014). In addition, several of the discovered genes, mainly the ones discovered in wild species, are linked with undesired traits. Discovery of genes in breeding lines and adapted germplasm, on the other hand, would more readily facilitate the use of such genes in breeding programs. Of various strategies implemented to map causative loci for segregating traits, nested association mapping (NAM) is used to map loci in a multi-cross mating design where one common parent is shared among all other ‘founder’ parents. Also known as joint mapping or joint linkage mapping, this strategy uses the benefits of both linkage mapping and linkage disequilibrium mapping to provide higher mapping power and resolution. Briefly, the strategy involves crossing several founder lines to a single common parent to generate segregating progenies in each population. The genetic background is normalized by virtue of having a common parent, which allows mapping of segregating alleles in different populations with reference to common-parent specific alleles (Blanc et al. 2006; Yu et al. 2008). Therefore, joint mapping helps to minimize problems that may arise due to genetic heterogeneity, different environmental effects, or simply experimental and sampling differences as all populations are connected by a common parent. The NAM design is also able to detect QTL with various effect sizes, including rare alleles, because of its higher statistical power. The efficacy of a NAM design in mapping important QTL have been demonstrated in recent studies in maize (Buckler et al. 2009; McMullen et al. 2009), Arabidopsis (Buckler and Gore 2007), and in a few other crop species (see Guo et al. 2010). 87

One key requirement of a sound mapping study is the abundance of high quality markers for genotyping the population. High marker coverage of the genome allows high resolution and accurate mapping of causative loci. Single nucleotide polymorphism (SNP) markers are preferred over other marker systems for genotyping because of their abundance, low cost per data point, and utility to high-throughput technologies. However, genotyping of populations using SNPs from pre-designed assays (or chips) is known to introduce founder ascertainment bias and are known to result in less accurate and biased data analysis (Albrechtsen et al. 2010; Heslot et al. 2013). One alternative to obtain a high number of high-throughput SNPs without ascertainment bias is the complexityreduced genome sequencing approach called genotyping by sequencing (GBS). GBS uses restriction enzymes for targeted complexity reduction of genomes followed by next-gen sequencing of multiplexed samples (Elshire et al. 2011; Poland et al. 2012b). The millions of reads thus obtained are used to discover SNPs, thus allowing the discovery of high quality population-specific SNPs for genomic studies. This approach is also appealing because of the low cost per sample, relatively faster turnaround time, and malleability in terms of sequence manipulation and data mining. This technique has been successfully used in wheat studies to obtain de novo genetic maps (Saintenac et al. 2013a), and in barley to map alleles influencing plant height (Liu et al. 2014). In this study, we use ten biparental spring wheat RIL populations to map stem rust resistance QTL. The objectives of this study were to conduct a genome-wide scan for QTL in a joint analysis of all ten populations, linked by a common parent. We use GBS markers for the first time to conduct a NAM analysis to dissect the relationship between 88

the genetics of a large mapping population and their resistance to African and North American Pgt races. All male parents used in the crosses have been released as commercial wheat varieties in their target areas. As such, we expect that the significant markers identified in this study will provide higher value and incentive towards introgression of the detected QTL for resistance breeding in areas that are hotspots and/or are vulnerable to the stem rust disease of wheat.

89

MATERIALS AND METHODS

Plant Material Nine Kenyan spring wheat lines (‘Fahari’, ‘Gem’, ‘Kudu’, ‘Kulungu’, ‘Ngiri’, ‘Paka’, ‘Pasa’, ‘Popo’, and ‘Romany’) that were released as varieties in the 1960s, 70s, and the 80s were selected for crossing based on the high level of APR exhibited by these lines during screening in the Njoro stem rust nursery in Kenya (Table 1). ‘Ada’, a recent hard red spring wheat variety released by the University of Minnesota (Anderson et al. 2007), also exhibits moderate resistance to African Pgt races in the Ug99 lineage and a high level of resistance against North American Pgt races (Table 1). These ten lines were crossed to the universal susceptible line ‘LMPG-6’ to develop recombinant inbred line populations (F6 or more inbred) via the single seed descent method at the University of Minnesota. The number of inbred lines in the ten populations ranged from 55 to 110 (Table 1). The pedigree information, known genes based on marker screening, and stem rust reactions of each parent are provided in Table 1.

Disease Phenotyping The ten RIL populations, comprised of 852 lines, were evaluated for adult plant reaction to stem rust in four environments: St. Paul, MN, USA during May-August 2012 & 2013 (referred as StP12 and StP13 in the text); South Africa during October 2012 – January 2013 (referred as SA12 in the text); and Njoro, Kenya during May-October 2013 (referred as Ken13 in the text).

90

In the Njoro nursery, lines were planted in an augmented design with 1 check, ‘Red Bobs’. Each line was sown in double 70 cm long rows, 20 cm apart. On each side of the plot, and in the middle of the plots, a twin-row of susceptible spreader wheat cultivar ‘Cacuke’ was sown. The field was also surrounded by a border of several spreader rows comprised of susceptible wheat varieties which were artificially inoculated using a bulk inoculum of Pgt urediniospores collected at the Njoro field site; however wheat stem rust differential lines with known stem rust resistance genes indicated that the predominant, if not only, race present in the nursery since 2008 was race TTKST (avirulence/virulence formula on the wheat stem rust differential panel: Sr36, SrTmp/Sr5, Sr6, Sr7b, Sr8a, Sr9a, Sr9b, Sr9d, Sr9e, Sr9g, Sr10, Sr11, Sr17, Sr21, Sr24, Sr30, Sr31, Sr38, SrMcN) (Njau et al. 2010). In the St. Paul 2012 environment, lines were planted in 2 m long single rows with 20 cm between the rows. The populations were planted in an augmented design with 4 check varieties ‘Oklee’ (Anderson et al. 2005), ‘Thatcher’ (Hayes et al. 1936), ‘Tom’ (Anderson et al. 2012), and ‘Verde’ (Busch et al. 1996) planted after every 30 entries. In the St. Paul 2013 environment, lines were planted in hill-plots with 20 cm distance between the hills. The same checks were planted in the same manner as in the 2012 season. In both environments, the lines were surrounded by a mixture of susceptible lines ‘Morocco’, ‘Thatcher’, ‘Max’, and ‘Little Club’ planted perpendicular to the lines on all sides. To initiate disease, spreader rows were syringe-injected with a mixture of North American stem rust races MCCFC (isolate 59KS19), QFCSC (isolate 03ND76C), QTHJC (isolate 75ND717C), RCRSC (isolate 77ND82A), RKQQC (isolate 99KS76A), 91

and TPMKC (isolate 74MN1409) at the jointing stage. The spreader rows were sprayed with a bulked mixture of Pgt races suspended in a light mineral oil suspension using an Ulva+ sprayer (Micron Sprayers Ltd., Bromyard, UK) after heading stage. In South Africa, populations were planted in Cedara of KwaZulu-Natal Province as an augmented design with the check lines ‘Morocco’ and ‘Kariega’ planted after every 50 entries. Seeds for each line were planted in hill plots with 30 cm between plots in a row and 60 cm between rows. To initiate the disease, spreader rows containing ‘Morocco’ and ‘McNair’ were inoculated with a mixture of the Pgt races TTKSF and PTKST of the Ug99 race group using an ultra low volume sprayer twice in the season – once during the booting stage, and again at flowering stage. Additionally, all 11 parent lines were inoculated with race TTKSK at seedling stages to postulate the presence/absence of major genes providing resistance to this widely virulent Pgt race. Disease inoculation and phenotyping procedures were carried out as described by Rouse et al. (2012).

Statistical Analysis Mixed linear models were fitted to each environment to assess family and line within family genotypic effects, and also the effects of heading date and growth stages of the lines. Factors explaining significant amounts of variation were retained in the model based on the likelihood ratio tests calculated in ‘lme4’ package 3.0.3 in R (R Development Core Team 2013, http://www.r-project.org/). Phenotypic data for each line was adjusted based on either the number of days to heading (St. Paul data) or growth stage (Kenya data) by estimating the effect of days to heading or growth stages on stem 92

rust severity for each line within each RIL population, and subtracting the estimate from rust severity for each line. The number of days to heading was measured as the day after planting when half the spikes in the plot fully emerged above the flag leaf. Growth stages of each line in Kenya were determined mainly by assessing grain development stages such as watery, milky, soft dough, and hard dough; and also for stages of booting and flowering, as explained by Zadoks et al. (1974). The phenotypic scale used in South Africa was different than the other locations as the disease was evaluated on a quantitative scale of 0 to 10 representing highly resistant (scores 0-2), resistant (score 3), moderate resistance (scores 4-5), moderate susceptible (scores 6-7), susceptible (score 8), and highly susceptible (scores 9-10) (Tsilo et al. 2014). Therefore, all data sets were fitted into a mixed model with environments as fixed effects and lines as random effects to correct for data distortion due to trial effects. Using this model, best linear unbiased predictors (BLUPs) for each line were predicted from the combined analysis model using SAS 9.1, from which final adjusted trait values were obtained.

Genotyping, SNP Discovery & Imputation of Missing Alleles All lines used in the study were genotyped using the genotyping by sequencing (GBS) approach modified after Elshire et al. (2011). Genomic DNA was extracted from ground leaf tissue of F6:7 RILs and parent seedlings using the BioSprint 96 DNA Plant Kit (QIAGEN, Valencia, CA). Extracted DNA was quantified using the picogreen assay and diluted to 20ng/µl. The restriction enzymes PstI and MspI were used to generate double-digested complexity-reduced DNA libraries following Poland et al. (2012b). The PstI restriction overhang was preceded by barcoded forward adapters, whereas the MspI 93

overhang was fitted with a common Y-adapter. This design prevented either PstI-PstI or MspI-MspI fragments from amplifying, thereby yielding sequences of only the PstI-MspI fragments, producing a uniform library. Few modifications were introduced into the protocol, mainly: 1) each sample was ligated to two unique barcodes to minimize sequencing bias of reads with certain barcodes; and 2) the concentrations of the barcode adapter and the common adapter were increased to 0.1 µM and 50 µM, respectively, in an attempt to capture most of the digested DNA fragments. Ten 96-plex libraries were generated, with each parent repeated at least six times to obtain higher read coverage of parental alleles. Each library was sequenced in one lane of Illumina HiSeq 2000, generating 100 bp single-end sequences. The Universal Network Enabled Analysis Kit (UNEAK) was used to call de novo SNPs across the populations using the parameters -c 10 –e 0.025 (Lu et al. 2013). Chromosome locations of the SNPs were obtained by aligning the reads containing SNPs to the wheat chromosome survey sequences (CSS), using the reads as query sequences in a local ‘blastn’ search. The wheat CSS sequences were obtained by assembling reads obtained from sequencing flow-sorted wheat chromosomes from ‘Chinese Spring’ (International Wheat Genome Sequencing Consortium, http://wheaturgi.versailles.inra.fr/Seq-Repository/). To ensure that correct SNPs were obtained, the following filters were applied to the blast results: 1) each read was allowed only one alignment to a unique chromosomal location; 2) full length alignment of each read was required; and 3) either one base mismatch or one gap between the query sequence and the wheat CSS was permitted, assuming some level of sequence 94

polymorphism existed between Chinese Spring and lines in our study. Our stringency parameters are modeled after Wang et al. (2014) who used a similar approach to assign their 91,829 SNPs on the Infinium iSelect array to wheat chromosomes. The workflow of genotyping, map construction, data analysis, and genetic mapping in this study is presented in Figure 1. One unique property of the GBS approach is the generation of significant proportions of missing genotype data (Williams et al. 2010; Poland et al. 2012b; Fu et al. 2014). As such, construction of linkage maps with missing data is an arduous task. However, the NAM design of mapping populations allows for imputation of parental SNPs to the segregating progeny with high accuracy (Guo and Beavis 2011). As both parents were sequenced multiple times, the high-confidence allele calls between the parents can be used to impute missing haplotypes in the progeny. Hence, imputation of missing SNP data based on family relationship was carried out using principal component analysis (PCA) based imputation using the probabilistic PCA (ppca) algorithm in the freely available R package ‘pcaMethods’ (Stacklies et al. 2007). The ppca algorithm first assigns row average values to the missing values, and then uses the singular value decomposition of the SNP matrix to create orthogonal principal components. In turn, the PC values corresponding to the largest eigenvalues are used to reconstruct the missing SNP genotypes in the data matrix. The algorithm ‘ppca’ was chosen for its high imputation accuracy and efficiency in regards to the use of computational resources compared to other imputation algorithms of similar caliber (Moser et al. 2009; Fu 2014).

95

Map Construction and Joint QTL Mapping The genetic map was constructed using the program IciMapping 3.3 (Wang et al. 2012a) using a minimum logarithm of odds (LOD) value of 5.0. Genetic distances between the markers were calculated based on the Kosambi mapping function (Kosambi 1943). A search for QTL across all populations, and in each population was done using MCQTL 5.2.6 (Jourjon et al. 2005). Within MCQTL, the iterative QTL mapping (iQTLm) procedure (Charcosset et al. 2000) was implemented with walk distance of 5 cM to detect significant QTL. The iQTL method works in two steps where significant QTL are first scanned along chromosomes for their precise locations, followed by adding new QTL to or dropping QTL from the model based on Fisher’s test. Permutation analysis of the trait data was implemented using a resampling method as described by Churchill and Doerge (1994). Trait threshold was computed as described by Jourjon et al. (2005) to adapt multiple cross design using 1,000 permutation steps. Significant QTL were declared if the LOD threshold level was greater than the calculated value at α = 0.05. Epistasis effects among the detected QTL were also estimated, using the joint connected model in MCQTL 5.2.6, with LOD threshold calculated as described above. QTL mapping of each population was also carried out separately using the composite interval mapping (CIM) approach in MCQTL 5.2.6. For each detected QTL, the percent of phenotypic variance explained (R2) and allelic effects were also estimated.

96

RESULTS AND DISCUSSION

Genotyping We generated ten 96-plex GBS libraries representing 852 RILs from ten biparental populations. Each parent line was sequenced at least six times to obtain sequences with higher read depth for confidence in SNP-calling and imputation. As a result, the sequencing of each library on one lane of Illumina HiSeq 2000 generated a total of 1.5 billion 100 bp reads, with 154 million reads on average per lane. On average, 90% of the generated bases passed the Q30 filter with a median Q-score of 34.98. The reads were then filtered for having intact barcode sequences and a complete PstI overhang, which led to 73% of total reads assigned to each individual. This figure is comparable to recent GBS studies in wheat (Saintenac et al. 2013a) and barley (Liu et al. 2014) that follow the same library construction and sequencing protocols. This resulted in a read distribution per individual from 175,443 to 31,381,071 with a median of 1,225,681 reads per RIL. A total of 158,182,462 reads were obtained for the 11 parents used in the study, with a median value of 13,348,322 reads, and ranging from 7,951,649 (Kudu) to 31,381,071 of the common parent LMPG-6 which was replicated 10 times in the library (Appendix IV). The GBS approach has been used in species where reference sequences are available (Elshire et al. 2011; Liu et al. 2014), and in species where reference sequences are not available (Baxter et al. 2011; Saintenac et al. 2013a). Owing to the power in data analysis from the generated sequences, this approach has been successfully used in several genomic studies in non-model species as well as in species with high genome 97

complexity, such as that of hexaploid bread wheat. The large number of reads generated, as evident by the numbers reported here, enable the discovery of abundant polymorphic markers to be used in studies ranging from genetic mapping to population genomics. Regardless of the approach used to analyze the sequences and research questions asked, this method is efficient, generates better value to cost ratio than most other available genotyping systems, and therefore, is of high usability in investigating marker-trait associations (Chapter 3 of this thesis).

SNP Discovery and Linkage Mapping SNPs for which the SNP-containing reads matched uniquely to the WCSS were retained and passed through imputation steps. The imputed SNPs were compared against all 10 populations to remove inter-population common SNPs and retain only unique polymorphic SNPs. This led to identification of 992 SNPs that were used for construction of linkage maps and QTL mapping. Of the 992 high quality SNPs, 930 SNPs were assigned to 21 linkage groups with all 21 wheat chromosome represented. The number of markers per linkage group ranged from 1 (Chromosomes 3D, 7D) to 288 (Chromosome 6A). The total genetic distance covered by these groups was 4,977 cM, with one SNP marker placed at every 5 cM on average. The distribution of SNPs by linkage groups is shown in Figure 2. Linkage groups representing chromosomes 3B and 5B, were larger compared to other linkage groups, at 604 cM, and 696 cM, respectively. We suspect that the large amount of missing data in the genotype matrices representing these chromosomes led to this size inflation. The amounts of missing data on 3B and 5B genotype matrices were 18% and 39%, respectively. The remaining SNPs mapped to the 98

other 19 chromosomes together had less than 6% missing data. We decided to retain these SNPs as their removal would have resulted in insufficient number of SNPs for linkage group construction. Figure 2 also shows the read coverage per SNP, arranged by their positions on each linkage group. The median number of read coverage per SNP was 346, and ranged from 105 to 7,943. This range in read coverage per SNP is likely an attribute of uneven sampling of the genome during sequencing, either from non uniform representation of the samples, or from a large number of reads being discarded during our stringent sequence filtering and SNP calling procedures.

Population Characteristics and Stem Rust Reaction Population structure among the populations was determined by singular value decomposition of the genotype data using the ‘princomp’ package in R 3.0.3 (R Development Core Team 2013, http://www.r-project.org/). No significant structuring among the lines was observed (Appendix V), which is expected of NAM population design (Yu et al. 2008). The lack of population structure among the populations despite having diverse founder lines is an attractive aspect of the NAM design. As different founder lines are crossed to a common parent, shuffling of the parental genomes in the progeny normalizes allelic differences by virtue of the common parent. Therefore, population stratification is effectively controlled in the NAM design, which minimizes spurious associations that could arise from population structure. Because of the lack of structure, relationship matrices were not used as cofactors during the data analysis.

99

We phenotyped all ten RIL populations and the parent lines over two years at four environments where disease epidemics were artificially established by inoculating spreader lines with local Pgt races. Screening for APR to North American and African Pgt races showed that all male parents used in the cross were resistant to moderately resistant (Table 1). Some susceptible pustules were observed on the stem of ‘Ada’ in Kenya 2013 season, albeit at low severity (average of 11%). The susceptible parent ‘LMPG-6’ succumbed to high disease severity in all four environments and exhibited high susceptibility. Phenotyping of all populations was initiated once disease severity on ‘LMPG-6’ was at its maximum at each environment. All populations exhibited continuous disease distribution in most trials, allowing mapping of quantitative loci associated with disease resistance (Appendix V). The frequency of resistant lines was higher in the Kenya 2013 environment than any other, most likely due to the unusually cooler day and night temperatures at this site during the 2013 main season, which led to slower onset of disease than usual. Adjustment of phenotypic data was able to resolve this, as trait means among all environments were similar post-adjustment. Seedling screening of the parent lines with the race TTKSK showed that all parents were susceptible to this race (Table 1). The parent lines were screened for the presence of resistance genes effective against races of the Ug99 lineage for which diagnostic DNA markers have been developed (MAS Wheat 2014). While the results indicated that the lines ‘Gem’, and ‘Romany’ contain the gene Sr22 which is effective to all seven races of the Ug99 lineage (Singh et al. 2011), all lines were found to be susceptible to the Ug99 lineal race TTKSK during seedling screening, suggesting that 100

these lines lack any seedling gene (Table 1). The discrepancy between detection of Sr22 and the susceptibility of these two lines to race TTKSK has been discussed in detail in the ‘Comparison with Previously Reported Genes and QTL Conferring Resistance to Pgt’ section.

Stem Rust QTL Mapping We implemented the iterative QTL mapping (iQTLm) method to map QTL in all RIL populations providing APR to stem rust in four environments. Results of the joint QTL analysis of the NAM panel are presented in Figure 3 and Appendix IX99. We also mapped QTL in individual populations using the composite interval mapping (CIM) method, and the results from this analysis are summarized in Figure 4 and Appendix IV. The iQTLm approach discovered 27 QTL on 11 of 21 wheat chromosomes (Figure 3). This number of QTL detected in joint mapping is larger than that detected in any single-population QTL detection using the CIM approach (Appendix IV), although the QTL effects and R2 values are comparable between the two approaches. Our findings are in agreement with published NAM studies where additional QTL have been discovered in joint mapping compared to single-populations (Buckler et al. 2009; Negeri et al. 2011). Six of the 27 QTL were represented multiple times in more than one environment. The QTL QSr.umn-2A.1 and QSr.umn-4B.1 were common between Ken13 and Stp13 environments; QSr.umn-4A.1 between Ken13 and StP12; and QSr.umn-2B.2 (no relationship with QSr.umn-2B.2 in Chapter 2), QSr.umn-3B.4, and QSr.umn-3B.7 between StP12 and StP13 environments (Table 3, Appendix VI).

101

The 27 QTL detected by the NAM approach explained a wide range of phenotypic variation (R2) for adult plant resistance to stem rust of wheat, with the lowest R2 of 0.6% and the highest of 5.0%. These values indicate the absence of any large-effect QTL/gene contributing towards disease resistance. Rather, several QTL seem to contribute towards resistance in a cumulative fashion, as confirmed by the relatively small additive values in Figure 3. However, small population size can inflate phenotypic estimation variation (Eucharia 1990). As some populations in our study are small, the R2 values and additive effect estimates will have greater error vs. populations of larger size. Further phenotyping with larger populations is required to more accurately assess the R2 value contributed by the detected QTL. Interestingly, the additive effects of the QTL were similar to those that have been reported in stem rust QTL mapping studies carried out in biparental populations (Bhavani et al. 2011; Macharia 2013; Singh et al. 2013d). For each QTL, the parental alleles contributed different effects towards stem rust resistance, as exhibited by the distribution of total additive phenotypic effects of each parent, as shown in Figure 3. Overall, each parent contributed towards resistance in each environment. The positive values associated with the parent lines for specific SNPs in certain environments do not imply that the parents are not important sources of resistance. As the effects of all parental alleles are considered simultaneously during joint mapping, variations in disease resistance in each family, contributed by each parent, are expected to occur. Moreover, the choice of a parent for introgression of alleles for APR is also the function of several important factors defining the target environment such as frequency, and types of Pgt races, and disease pressure. While the contribution to disease 102

resistance from alleles with low additive values might be difficult to visually observe in the field, APR genes acting in an additive manner should elevate the resistance, thereby assisting in phenotypic selection. We did not detect epistasis within the populations, or in a joint setting with all populations combined. Recent studies have mixed results for detection of epistatic interactions between the mapped QTL as some studies report significant levels of interaction (Yu et al. 2012; Singh et al. 2013a; Singh et al. 2013b; Rouse et al. 2014) whereas other studies do not (Bhavani et al. 2011; Macharia 2013; Singh et al. 2013d). Taken together the number of QTL detected during our analysis and magnitude of the marker effects, it could be said that the nature of resistance to stem rust in these ten populations is polygenic, with several loci of minor to major effect acting in additive fashion. A point to consider is that the lack of epistasis should not deter a breeder from adoption of a QTL as long as it contributes to elevated APR, recognized mainly from its additive value. It should also be noted that the absence of a QTL effect in some environments does not necessarily mean that the locus had no effect on disease resistance. It is possible that the stringent threshold settings of the analysis filtered the QTL out from being detected in certain chromosomal regions, especially if they were slightly under the threshold value. This is likely to happen especially due to inadequate marker coverage in such regions. Another possible explanation is because environmental differences alter the expression of the genes, causing the loci to go undetected in some environments during analysis. The breeding history of the founder parents, in part, could also be responsible 103

for the observed differences in QTL x environment effects. As the varieties are selected for optimal performance in different environments, it is possible that certain loci are fixed while others are purged, producing a wide range of results during genome mapping. In our study, the distance of a SNP nearest to the reported QTL peak ranged from 0 cM to 20 cM, with an average value of 4 cM. Only four QTL were detected with SNPs further than 5 cM from the QTL peak position. The average extent of intra-chromosomal LD in spring wheat population is reported to be 20.8 cM (median value of 11.5 cM) and the rate of LD decay is about 6 cM (Chao et al. 2010). Hence, as most of the discovered significant markers close to the QTL peak, they can be expected to remain in LD with the causative loci and can provide high value in marker assisted selection for resistance breeding.

Comparison of Joint Mapping to Single-population QTL Mapping We were interested in knowing if the QTL detected during joint mapping could also be detected by conducting QTL mapping of each population separately. The CIM approach of QTL detection conducted separately on each RIL population found a total of 56 QTL on 17 chromosomes. No QTL were detected on chromosomes 1D, 3D and 4D, 5A. The population LMPG-6/Fahari had the largest number of QTL (15 QTL), and the populations LMPG-6/Kulungu and LMPG-6/Paka had the least number of QTL (1 QTL each) (Figure 4). No QTL were detected in LMPG-6/Ngiri population. No QTL were detected in all four environments in any individual RIL population, although seven populations had at least one QTL effective to Pgt races in three of the four environments.

104

The joint mapping approach detected QTL in all environments and on all chromosomes where QTL were detected by single-population QTL mapping. The singlepopulation mapping (upon summation of number QTL detected in all populations) and joint mapping methods detected the same number of QTL (11) in the Kenya 2013 environment. In all other environments, this number was higher in single-population mapping with 9 QTL detected in SA12, 14 QTL in StP12, and 21 QTL in StP13 relative to 2, 11, and 9 QTL detected in SA12, StP12, and StP13 environments, respectively, in joint mapping (Figure 4, Appendix IV). The percent of phenotypic variation (R2) explained by the QTL as well as the additive effects contributed by the parents between the two methods were different. The R2 value of a QTL detected in single-population mapping ranged from 0.1% to 24.6% whereas it ranged from 0.6% to 5% in joint mapping. The additive values of the parents calculated for each QTL in the joint mapping approach ranged from -33 to +19 whereas it ranged from -4 to +8 for QTL detected in individual populations. As QTL in joint mapping are declared based on their overall contribution to resistance (or susceptibility) across all 10 families, a ‘normalized’ value is calculated with the effect of each parent taken into account. In other words, a joint analysis estimates the allelic effects of a QTL by assuming that the allele from a founder parent is uniform across all populations (Rebai and Goffinet 1993; Blanc et al. 2006). This might not necessarily be true for each QTL detected, especially in case of linked genes, and among populations that differ in the number and effect of loci associated with the trait. In our study, via CIM mapping of single populations, we have found that not all populations share the same locus for stem rust resistance (Appendix IV). Hence, such an 105

assumption in the joint mapping model could be one of the main reasons why the joint mapping differs from individual mapping in estimation of the amount of total phenotypic variance and the allelic effect accounted by the significant SNPs. The two methods also detected a few QTL that were consistent in several disease environments. Eleven QTL distributed on eight chromosomes that were detected by the joint mapping approach were also detected by CIM in single populations (Table 3). The Pearson correlation value for the percent of phenotypic variance explained (R2) by the QTL detected between the two methods was moderately high at 0.55, implying the similarity between the QTL detected by the two methods. The biggest differences in the R2 value were observed in four QTL detected between the SA12 and StP12 environments where the values differed from 11.2% to 20.5%. Differences due to environmental conditions, inoculum load, and Pgt races present in each environment are probable to contribute to the differences in QTL detected in different environments. As shown in Table 3, three of the 11 QTL were detected in different environments between the two methods: the QTL detected on chromosome 3B in Ken13 and StP13 environments by joint mapping were detected in environments StP13 and Kenya 2013, respectively, in single-population mapping; and the QTL detected on 5D in the StP12 environment by joint mapping was detected in SA12 environment in single-population mapping. The difference of the environments where the QTL were detected suggests that these loci might be involved in providing resistance that is almost specific to the disease environments, and perhaps to the Pgt races. Validation of the QTL detected in both

106

approaches using the markers significantly associated with the QTL will help to further understand the relationship between the QTL detected in these two mapping approaches.

Comparison of Joint Mapping Results to Previously Reported Genes and QTL Conferring Resistance to Pgt Previous mapping studies have reported QTL conferring resistance to the Pgt races on the same chromosomes as detected in the present study. However, a joint mapping study with such objectives has not been carried out before. In our study, we detected one QTL on chromosome 2D, two QTL on 2A, and three QTL on 2B, of which the QTL QSr.umn-2A.1 was common between Ken13 and StP13 environments and the QTL QSr.umn-2B.2 was observed in both St. Paul environments (Figure 3). Several QTL conferring APR to Pgt races, including the Ug99 lineage races, on chromosomes 2A, 2B, and 2D have been detected in several spring wheat RIL populations and genome-wide association study (GWAS) panels as well as in durum wheat populations and GWAS panels (see Haile and Röder 2013; Yu et al. 2014). Several Sr genes are also located in the chromosomal regions where QTL for stem rust resistance were detected in this study. The genes Sr32 (located on 2A and/or 2B), and Sr28, Sr36, Sr39, Sr40, Sr47 (located on 2B) are effective to the races in the Ug99 lineage, but are unlikely to exist in our population based on screening for seedling resistance to TTKSK and screening of the parents using molecular markers. Several other genes such as, Sr6 (located on 2D), Sr9a, 9b, 9d, 9e (located on 2BL); Sr21 (located on 2AL), and Sr38 (located on 2AS) are ineffective against African races but are resistant to one or more North American Pgt 107

races (Rouse and Jin 2011; Zhang et al. 2014). Also located on the long arm of 2B, QSr.umn-2B.2 was detected in both StP environments, and QSr.umn-2B.3 was detected in StP12 environment. Kenyan lines are known to be the sources of many Sr genes discovered to date (McIntosh et al. 1995). Examples include Sr6, which is quite common in Kenyan lines (Knott 1962), and Sr9b, which was first observed in Kenyan lines (Knott and Anderson 1956). The line ‘Frontana’, present in the pedigree of the founder parent ‘Gem’, is known as one of the sources of Sr9b (McIntosh et al. 1995). Hence, our NAM population may possess these Sr genes, in addition to previously unidentified genes, and could be involved in providing resistance to North American Pgt races. The QTL QSr.umn-3B.4 detected on distal end of chromosome 3B may indicate the presence of Sr2, which is segregating in four populations: LMPG-6/Gem, LMPG6/Kudu, LMPG-6/Paka, and LMPG-6/Romany. The presence of Sr2 in ‘Gem’, ‘Kudu’, ‘Paka’, and ‘Romany’ was confirmed by marker screening (Table 1) as well as expression of the trait pseudo-black chaff (PBC) in St. Paul 2012 and Kenya 2013 environments on plant internodes (not observed in Romany). PBC is conditioned by the expression of the partially dominant gene Pbc causing dark coloring of the glumes and inter-nodal regions in an adult wheat plant, and is considered to be associated with the Sr2 gene (McFadden 1939; Sharp et al. 2001). The expression of PBC is highly dependent on the environment for its expression, and is also affected by the genetic background. The additive effect of the resistance allele at this QTL was observed to contribute towards resistance in populations (Figure 3) with the founder parents ‘Kudu’, ‘Paka’, and ‘Romany’, whereas the parent ‘Gem’ was found to contain an allele that 108

contributed towards susceptibility. Sr2 is an APR gene that provides broad-spectrum resistance to all known isolates of Pgt worldwide (McIntosh et al. 1995; Spielmeyer et al. 2003). Therefore, it is expected to be effective in all environments. As QSr.umn-3B.4 was observed in only two environments (StP12 and StP13) in our study, further field screening of the population is needed to confirm its identity relative to Sr2. We detected three QTL on the proximal end of 4B, which explained 0.6% to 4.0% of phenotypic variance observed in the four environments. Chromosome 4A contains the genes Sr7a, and 7b, and chromosome 4B hosts the genes Sr37. Sr37 is resistant to Ug99, and therefore is unlikely to be present in our population as the parents were seedlingsusceptible to Ug99. Both Sr7a and Sr7b are resistant to multiple N. American Pgt races (Roelfs and McVey 1979; Rouse and Jin 2011). Sr7a was first identified in several Kenyan wheat lines, (Knott and Anderson 1956; Knott 1962); and ‘Ngiri’ could have acquired this gene from the line ‘Manitou’ which is present in its pedigree (McIntosh et al. 1995). Thus, this gene may have been bred into one or more of the Kenyan varieties used to create our NAM population. Further screening of the population, preferably in single rust race nurseries, is needed to confirm this proposition. We detected three different QTL on 5B in Ken13, StP12 and StP13 environments that explained 1.6%, 1.8%, and 2.1%, respectively, of the variance observed for rust resistance in these environments. Two of these three QTL were also detected in singlepopulation mapping (Table 3). The QTL QSr.umn-5B.2 detected in StP12 environment was also detected in the same environment in LMPG-6/Popo population; and the QTL QSr.umn-5B.3 detected in StP13 environment was detected in the same environment in 109

the LMPG-6/Fahari population. No known Sr gene has been mapped to 5B, and the previously reported QTL effective to Pgt differ from those identified in our study in terms of their position and allelic effects. The QTL on 5D (QSr.umn-5D.2) detected in this study may be Sr30, which is known to exhibit resistance to North American Pgt races. Sr30 is ineffective against African races such as TTKSK at both seedling and adult plant stages (Mago et al. 2011a; Yu et al. 2014), yet exhibits seedling resistance to a few North American Pgt races (Rouse and Jin 2011) and all-stage resistance to Australian Pgt pathotypes (Bariana et al. 2001; Kaur et al. 2009). The total phenotypic variance explained by Sr30 was high (≥ 20%) in the study by Kaur et al. (2009) relative to that explained by QSr.umn-5D.2 (2%). Differences among the Pgt races used to initiate the disease, environments used for APR screening, and the genetic background of the parental lines could be the probable factors causing this variation. The QTL QSr.umn-6A, detected only in the Ken13 environment, explained 2.6% of phenotypic variance in this environment. QTL effective against races in the Ug99 lineage have been detected on 6A previously, including Sr26 in common wheat (Prins et al. 2011a; Yu et al. 2011); and Sr13 in durum wheat (Triticum durum Desf.) (Pozniak et al. 2008; Letta et al. 2013). Sr26 is effective against all races of the Ug99 lineage (Mago et al. 2011a), whereas all the NAM parents were susceptible to race TTKSK (Table 1). Also, as Sr13 is not prevalent in common wheat (McIntosh et al. 1995; Yu et al. 2014), it is unlikely that either of these genes is present in our population. Chromosome 6A also contains the gene Sr52, which was recently introgressed into hexaploid wheat from its 110

diploid relative Dasypyrum villosum, and is resistant to TTKSK (Qi et al. 2011). Being a novel gene introduced to wheat via translocation, Sr52 is not currently used in breeding programs (Yu et al. 2014), and therefore is not expected to be present in old Kenyan varieties used in this study. It is possible that QSr.umn-6A is a novel source of resistance to the Ug99 group of races. Further tests are needed to confirm this hypothesis. Pozniak et al. (2008), in their association mapping study comprised of durum wheat lines, report a QTL providing APR to the Ug99 lineage races located at 83 cM on 7A. The gene Sr22 is also located on 7A. This gene was detected in lines ‘Gem’ and ‘Romany’ during our marker screening (Table 1). While this may suggest the presence of Sr22 in our population, it should be noted that Sr22 is effective to TTKSK at the seedling stage (Mago et al. 2011a), which we did not observe in either ‘Gem’ or ‘Romany’ during our tests for seedling resistance. We suspect that either the markers we used amplified a null allele, or that the markers are not completely diagnostic, as confirmed by Olson et al. (2010a), whose recommended markers were used for screening of the gene. The QTL QSr.umn-7B detected in the SA12 environment explained 4.1% of the observed phenotypic variance, and is likely novel. The only gene mapped to 7B is Sr17, which is ineffective to races of the Ug99 lineage as well as to the North American Pgt races (Mago et al. 2011a; Rouse and Jin 2011). Previously reported QTL on chromosome 7B that confer resistance to the Ug99 group of races include the QTL linked to the microsatellite markers wPt-0318 (Yu et al. 2012) and cfa-2040 (Pozniak et al. 2008). Additional research, either jointly or in individual populations, is necessary to further

111

understand the relationship between discussed All-stage genes with the QTL detected in our study.

112

CONCLUSION In this study, we present results of a nested association mapping approach, an effective mapping strategy that utilizes the mapping power of several RIL populations. In addition, by using a statistical procedure designed for NAM, we estimate allele effects by using the common parent as a common genomic reference. Joint QTL mapping for stem rust resistance resulted in detection of a large number of QTL with small to large effects. Epistatic effects among the detected loci did not have significant contributions to stem rust resistance, suggesting that the differences between the populations in different environments are largely due to additive effects of several QTL. Also, by sequencing complexity-reduced genomes of the whole NAM panel, we obtained population-specific de novo SNPs, and show their usability in mapping QTL effective against stem rust of wheat. While validation of the detected loci is required to confirm the significant markers, GBS offers an efficient and economical approach for genome mapping and genomewide studies. However, multiple rounds of sequencing is recommended to obtain enough reads to minimize the complications that could arise from having substantial amounts of missing data. Sequencing the parent lines at higher read depth helps towards more accurate imputation of missing haplotypes. The genomic regions associated with stem rust resistance identified in this study will be useful to breeders in introgression of resistance to stem rust of wheat, including the widely virulent African stem rust races. Validation of our de novo markers by fine mapping of the regions and/or in marker assisted breeding is needed to confirm the locations of the reported QTL.

113

Table 1: Origin, pedigree, and stem rust reaction of parent lines used to develop the NAM population.

Parent

Ada Fahari Gem Kudu Kulungu Ngiri Paka Pasa Popo Romany LMPG-6

a

a

Origin

USA (2007) Kenya (1977) Kenya (1964) Kenya (1966) Kenya (1982) Kenya (1979) Kenya (1975) Kenya (1989) Kenya (1982) Kenya (1966) Canada (1990)

Pedigree

a

Sr markers

b

c

TTKSK

d

Field reaction USA

Africa

e

RILs

SBY189H/‘2375’

3+

22.3

11.3

71

TOBARI-66/3/SRPC-527-67//CI-8154/2*FROCOR

33+

11.7

3.7

90

BT908/FRONTANA//CAJEME 54

Sr2, Sr22

3-

10.2

1.0

97

KENYA-131/KENYA-184-P

Sr2

3+

19.8

3.7

80

ON/TR207/3/CNO//SN64/4/KTM

33+

22.5

0.3

59

SANTACATALINA/3/MANITOU/4/2*TOBARI-66

33+

10.2

2.3

52

3+

16.5

3.7

104

BUCK BUCK/CHAT

3+

8.6

5.3

93

KLEIN-ATLAS/TOBARI66//CENTRIFEN/3/BLUEBIRD/4/KENYA-FAHARI

3+

5.0

2.3

97

3+

11.1

5.3

109

4

64.9

25.0

-

WISCONSIN-245/II-50-17//CI-8154/2*TOBARI-66

COLOTANA 261-51/YAKTANA 54A

Sr2

Sr2, Sr22

LITTLE-CLUB//PRELUDE*8/MARQUIS/3/GABO

Information obtained from Njau et al. (2009), Macharia (2013), Anderson et al (2007), and Knott (1990). Year (in

parenthesis) indicates the year the line was released or published 114

b

Putative genes present in the NAM parents based on marker screening. Marker screening was conducted using diagnostic

markers for genes resistant to Ug99 races, as catalogued in MAS Wheat (2014) http://maswheat.ucdavis.edu/. The lines ‘Gem’ and ‘Romany’ appear positive for the presence of Sr22 but are susceptible to Ug99, implying that the markers used for screening are not diagnostic of the gene (see ‘Results & Discussion’ section for detailed explanation) c

Seedling screening of the parent lines with race TTKSK (isolate ‘04KEN156/04’)

d

For USA environment, the rust response of parent lines were averaged from StP12 and StP13 environments. For Africa

environment, only Ken13 data is shown. Mean severity reactions (%) are shown for both sites e

The number of progenies in the population obtained by crossing the parent line to LMPG-6

115

Table 2: Common quantitative trait loci (QTL) detected between the iterative QTL mapping (iQTLm) in nested association mapping of 10 RIL populations, and composite interval mapping (CIM) methods in individual populations for stem rust adult plant resistance in four environments. Env

Chra Left SNP Right SNP Posb

d

LODc PVE (%)

Env

Chra Left SNP Right SNP Pos b

d

LODc PVE (%) Population

SNP4

SNP813

723.8

2.9

1.9

Ken13 3B

SNP885

SNP790

684.5

3.5

4A

SNP371

SNP375

164.3

2.5

1.5

4A

SNP371

SNP375

164.3

3.3

2.7 LM PG-6/Kudu 1.2 LM PG-6/Kudu

6A

SNP935

SNP425

210.9

4.2

2.6

6A

SNP935

SNP425

210.3

2.6

10.2 LM PG-6/Romany

4B

SNP177

SNP203

358.8

7.3

4.0

4B

SNP177

SNP203

358.8

4.7

7B

SNP268

SNP337

43.0

7.6

4.1

5D

SNP714

SNP481

143.0

3.7

17.4 LM PG-6/Pasa 15.6 LM PG-6/Gem

StP12 2B

SNP294

SNP970

439.9

6.1

3.8

7B

SNP268

SNP337

43.0

7.6

24.6 LM PG-6/Romany

3B

SNP320

SNP359

811.0

8.3

5.0

StP12 2B

SNP294

SNP970

439.9

3.0

4A

SNP371

SNP375

164.9

2.5

1.5

5B

SNP142

SNP784

629.2

3.1

5.7 LM PG-6/Gem 12.9 LM PG-6/Popo

5B

SNP142

SNP784

624.2

2.8

1.8

StP13 2B

SNP294

SNP970

429.9

2.7

5D

SNP714

SNP481

138.0

3.0

1.9

2B

SNP294

SNP970

434.9

3.0

StP13 2B

SNP294

SNP970

434.9

3.2

2.3

3B

SNP4

SNP813

713.8

3.7

3B

SNP885

SNP790

684.5

3.6

2.6

3B

SNP320

SNP359

806.0

2.5

3.4 LM PG-6/Ada 1.7 LM PG-6/Gem

3B

SNP320

SNP359

811.0

3.5

2.6

5B

SNP141

SNP765

575.9

2.6

1.2 LM PG-6/Gem

5B

SNP141

SNP765

575.9

2.9

2.1

Ken13 3B

SA12

SA12

a

Chromosome location of the QTL

b

Position (centiMorgan) of the detected QTL peak in Chromosome ‘Chrom’

c

Logarithm of odds scores for the QTL detected at position ‘Pos’, based on joint mapping

d

Percentage of phenotypic variation explained by the observed QTL, based on joint mapping 116

12.0 LM PG-6/Pasa 6.5 LM PG-6/Gem

Figure 1: Schematic workflow of the study. See Materials & Methods for detailed explanation of each step.

117

Figure 2: Distribution of SNPs by linkage groups (X-axis) and read depth (Y-axis) obtained from GBS approach. In each linkage group, SNPs are sorted by position. The numbers of SNPs in each linkage group are shown in parenthesis next to the linkage group. The dotted line represents the average SNP read coverage (546 reads).

118

Figure 3: Heat map of additive effect estimates of alleles contributed by the 10 founder lines at the QTL for resistance to Pgt races. QTL (columns) are named according to McIntosh et al. (2003); the allelic effect estimates for each founder allele (rows) are color coded by increments in the allelic effect estimate (legend); each block represents the environments where the QTL were detected, as labeled. Chrom = Chromosome location of the QTL; Pos = Position (centiMorgan) of the detected QTL peak in Chromosome ‘Chrom’; LOD = Logarithm of odds scores for the QTL detected at position ‘Pos’, based on joint mapping; PVE = Percentage of phenotypic variation explained by the observed QTL, based on joint mapping. 119

Ken13 12

SA12

StP12

StP13

11

11 11

Number of QTL

10

9

8

7

6

5 4

4 2 0

3 3 2

2

2

2

1 0

Ada

0

Fahari

3 2

1 0

Gem

1 0

Kudu

0 0 0

1 0 0 0 0

Kulungu

Ngiri

0 0

1 0

Paka

0

Pasa

2

2 1 1

2

2

1 0

Popo

0

Romany

Parent Lines

Figure 4: Frequency distribution of number of QTL discovered in different environments from composite interval mapping (CIM) of each RIL population and iterative QTL mapping (iQTLm) in joint mapping of all populations combined. Numbers next to the bars represent the number of QTL detected in all environments respective to the parent line. The environments Saint Paul 2012, South Africa 2012, Saint Paul 2013, and Kenya 2013 are abbreviated as StP12, SA12, StP13, and Ken13, respectively.

120

Joint

REFERENCES Albrechtsen A, Nielsen FC, Nielsen R (2010) Ascertainment biases in SNP chips affect measures of population divergence. Molecular Biology and Evolution 27:2534-2547 Allen AM, Barker GLA, Berry ST, Coghill JA, Gwilliam R, Kirby S, Robinson P, Brenchley RC, D’Amore R, McKenzie N, Waite D, Hall A, Bevan M, Hall N, Edwards KJ (2011) Transcript-specific, single-nucleotide polymorphism discovery and linkage analysis in hexaploid bread wheat (Triticum aestivum L.). Plant Biotechnology Journal 9:1086-1099 Anderson JA, Busch RH, McVey DV, Kolmer JA, Jin Y, Linkert GL, Wieserma JV, DillMacky R, Wieserma JJ, Hareland GA (2007) Registration of ‘Ada’ wheat. Crop Science 47:434-435 Anderson JA, Busch RH, Mcvey DV, Kolmer JA, Linkert GL, Wiersma JV, Dill-Macky R, Wiersma JJ, Hareland GA (2005) Registration of ‘Oklee’ wheat. Crop Science 45:784-785 Anderson JA, Linkert GL, Busch RH, Wiersma JJ, Kolmer JA, Jin Y, Dill-Macky R, Wiersma JV, Hareland GA, McVey DV (2009) Registration of ‘RB07’ wheat. J Plant Reg 3:175-180 Anderson JA, Ogihara Y, Sorrells ME, Tanksley SD (1992) Development of a chromosomal arm map for wheat based on RFLP markers. Theoretical and Applied Genetics 83:1035-1043 Anderson JA, Wiersma JJ, Linkert GL, Kolmer JA, Jin Y, Dill-Macky R, Wiersma JV, Hareland GA, Busch RH (2012) Registration of ‘Tom’ wheat. Journal of Plant Registrations 6:180-185 Arthur JC, Kern FD, Orton CR, Fromme FD, Jackson HS, Mains EB, Bisby GR (1929) The Plant Rusts (Uredinales). J. Wiley & sons, Inc, New York Ayliffe M, Singh R, Lagudah E (2008) Durable resistance to wheat stem rust needed. Current Opinion in Plant Biology 11:187-192 Bansal U, Bariana H, Wong D, Randhawa M, Wicker T, Hayden M, Keller B (2014) Molecular mapping of an adult plant stem rust resistance gene Sr56 in winter wheat cultivar Arina. Theor Appl Genet 127:1441-1448 Bansal UK, Bossolini E, Miah H, Keller B, Park RF, Bariana HS (2008) Genetic mapping of seedling and adult plant stem rust resistance in two European winter wheat cultivars. Euphytica 164:821-828 121

Bariana HS, Cupitt CF, Warburton T (2001) Diversity of resistance to rust diseases in Australian wheats in 1999 and 2000 crop seasons. 10th Assembly of Wheat Breeding Society of Australia, Mildura, Australia, pp 233-236 Bariana HS, Kailasapillai S, Brown GN, Sharp PJ (1998) Marker assisted identification of Sr2 in The National Cereal Rust Control Program in Australia. In: Slinkard AE (ed) Proceedings of the 8th International Wheat Genetics Symposium. University Extension Press, University of Saskatchewan, Saskatoon, Canada, pp 78-80 Bariana HS, McIntosh RA (1993) Cytogenetic studies in wheat. XV. Location of rust resistance genes in VPM1 and their genetic linkage with other disease resistance genes in chromosome 2A. Genome 36:476-482 Baxter SW, Davey JW, Johnston JS, Shelton AM, Heckel DG, Jiggins CD, Blaxter ML (2011) Linkage mapping and comparative genomics using next-generation RAD sequencing of a non-model organism. PLoS ONE 6:e19315 Bhavani S, Singh RP, Argillier O, Huerta-Espino J, Singh S, Njau P, Brun S, Lacam S, Desmouceaux N (2011) Mapping durable adult plant stem rust resistance to the race Ug99 group in six CIMMYT wheats. In: McIntosh RA (ed) BGRI 2011 Technical Workshop, Borlaug Global Rust Initiative, St Paul, MN, USA, pp 43-53 Bidwell PW, Falconer JI (1925) History of agriculture in the northern United States, 1620-1880. Carnegie Institute, Washington DC, USA Blanc G, Charcosset A, Mangin B, Gallais A, Moreau L (2006) Connected populations for detecting quantitative trait loci and testing for epistasis: an application in maize. Theor Appl Genet 113:206-224 Bonnett DG, Rebetzke GJ, Spielmeyer W (2005) Strategies for efficient implementation of molecular markers in wheat breeding. Mol Breeding 15:75-85 Botstein D, White RL, Skolnick M, Davis RW (1980) Construction of a genetic linkage map in man using restriction fragment length polymorphisms. American journal of human genetics 32:314-331 Browning JA (1979) Genetic protective mechanisms of plant-pathogen populations: their coevolution and use in breeding for resistance. In: Harris MK (ed) Biology and Breeding for Resistance. Texas A&M University, College Station, TX, USA, pp 5275 Buckler E, Gore M (2007) An Arabidopsis haplotype map takes root. Nat Genet 39:10561057 Buckler ES, Holland JB, Bradbury PJ, Acharya CB, Brown PJ, Browne C, Ersoz E, FlintGarcia S, Garcia A, Glaubitz JC, Goodman MM, Harjes C, Guill K, Kroon DE, 122

Larsson S, Lepak NK, Li H, Mitchell SE, Pressoir G, Peiffer JA, Rosas MO, Rocheford TR, Romay MC, Romero S, Salvo S, Villeda HS, Sofia da Silva H, Sun Q, Tian F, Upadyayula N, Ware D, Yates H, Yu J, Zhang Z, Kresovich S, McMullen MD (2009) The genetic architecture of maize flowering time. Science 325:714-718 Burdon JJ, Barrett LG, Rebetzke G, Thrall PH (2014) Guiding deployment of resistance in cereals using evolutionary principles. Evolutionary Applications 7:609–624 Burton GH (1928) Report of plant breeder, Department of Agriculture, Kenya Busch RH, McVey DV, Linkert GL, Wiersma JV, Warnes DO, Wilcoxson RD, Hareland GA, Edwards I, Schmidt H (1996) Registration of ‘Verde’ wheat. Crop Science 36:1418 Butler FC (1948) Stem rust of wheat: the value of resistant varieties. Agricultural Gazette of New South Wales 59:511–514 Caldwell RM (1968) Breeding for general and/or specific plant disease resistance. In: Findlay KW, Shepherd KW (eds) Proceedings of the 3rd Int Wheat Genet Symposium. Australian Academy of Science, Canberra, Australia, pp 263-272 Carleton MA (1905) Lessons from the grain-rust epidemic of 1904. Farmers’ Bulletin 219:1–24 Carter AH, Chen XM, Garland-Campbell K, Kidwell KK (2009) Identifying QTL for high-temperature adult-plant resistance to stripe rust (Puccinia striiformis f. sp. tritici) in the spring wheat (Triticum aestivum L.) cultivar ‘Louise’. Theoretical and Applied Genetics 119:1119-1128 Cavanagh CR, Chao S, Wang S, Huang BE, Stephen S, Kiani S, Forrest K, Saintenac C, Brown-Guedira GL, Akhunova A, See D, Bai G, Pumphrey M, Tomar L, Wong D, Kong S, Reynolds M, da Silva ML, Bockelman H, Talbert L, Anderson JA, Dreisigacker S, Baenziger S, Carter A, Korzun V, Morrell PL, Dubcovsky J, Morell MK, Sorrells ME, Hayden MJ, Akhunov E (2013) Genome-wide comparative diversity uncovers multiple targets of selection for improvement in hexaploid wheat landraces and cultivars. Proceedings of the National Academy of Sciences doi: 101073/pnas1217133110 Chao S, Dubcovsky J, Dvorak J, Luo M-C, Baenziger S, Matnyazov R, Clark D, Talbert L, Anderson J, Dreisigacker S, Glover K, Chen J, Campbell K, Bruckner P, Rudd J, Haley S, Carver B, Perry S, Sorrells M, Akhunov E (2010) Population- and genomespecific patterns of linkage disequilibrium and SNP variation in spring and winter wheat (Triticum aestivum L.). BMC Genomics 11:727

123

Chao S, Sharp PJ, Worland AJ, Warham EJ, Koebner RMD, Gale MD (1989) RFLPbased genetic maps of wheat homoeologous group 7 chromosomes. Theoretical and Applied Genetics 78:495-504 Chao S, Zhang W, Akhunov E, Sherman J, Ma Y, Luo M-C, Dubcovsky J (2009) Analysis of gene-derived SNP marker polymorphism in US wheat (Triticum aestivum L.) cultivars. Mol Breeding 23:23-33 Chao S, Zhang W, Dubcovsky J, Sorrells M (2007) Evaluation of genetic diversity and genome-wide linkage disequilibrium among U.S. wheat (Triticum aestivum L.) germplasm representing different market classes. Crop Science 47:1018-1030 Charcosset A, Mangin B, Moreau L, Combes L, Jourjon M-F (2000) Heterosis in maize investigated using connected RIL populations. Quantitative genetics and breeding methods: The way ahead. INRA Editions, Paris, France, pp 89–98 Chester KS, Gilbert FA, Hay RE, Newton N (1951) Cereal rusts: epidemiology, losses, and control. Battellc Memorial Institute, Columbus, Ohio Churchill GA, Doerge RW (1994) Empirical threshold values for quantitative trait mapping. Genetics 138:963-971 Cook, Hims, Vaughan (1999) Effects of fungicide spray timing on winter wheat disease control. Plant Pathology 48:33-50 de Bary A (1853) Investigations of the brand fungi and the diseases of plants caused by them with reference to grain and other useful plants [translated from German by Heffner RMS, Arny DC, Moore JD Phytopathological Classics 11; American Phytopathological Society, St. Paul, Minnesota, USA (1969)] Devey ME, Hart GE (1993) Chromosomal localization of intergenomic RFLP loci in hexaploid wheat. Genome 36:913-918 Deynze AEV, Dubcovsky J, Gill KS, Nelson JC, Sorrells ME, Dvořák J, Gill BS, Lagudah ES, McCouch SR, Appels R (1995) Molecular-genetic maps for group 1 chromosomes of Triticeae species and their relation to chromosomes in rice and oat. Genome 38:45-59 Dubin HJ, Brennan JP (2009) Combating stem and leaf rust of wheat: historical perspective, impacts, and lessons learned. International Food Policy Research Institute Discussion Paper:1-64 Dyck PL (1992) Transfer of a gene for stem rust resistance from Triticum araraticum to hexaploid wheat. Genome 35:788-792

124

Elshire RJ, Glaubitz JC, Sun Q, Poland JA, Kawamoto K, Buckler ES, Mitchell SE (2011) A robust, simple genotyping-by-sequencing (GBS) approach for high diversity species. PLoS ONE 6:e19379 Eucharia EN-O (1990) Health research design and methodology. CRC Press, FL, USA Evanega SD, Singh RP, Coffman R, Pumphrey MO (2014) The Borlaug Global Rust Initiative: reducing the genetic vulnerability of wheat to rust. In: Tuberosa R, Graner A, Frison E (eds) Genomics of Plant Genetic Resources. Springer Netherlands, pp 317-331 Fantham E (1998) Ovid: fasti book IV. Cambridge University Press, UK Faris JD, Haen KM, Gill BS (2000) Saturation mapping of a gene-rich recombination hot spot region in wheat. Genetics 154:823-835 Ferdosi M, Kinghorn B, van der Werf J, Lee S, Gondro C (2014) hsphase: an R package for pedigree reconstruction, detection of recombination events, phasing and imputation of half-sib family groups. BMC Bioinformatics 15:172 Fontana F (1767) Observations on the rust of grain. American Phytopathological Society, St Paul, Minnesota, USA Frascaroli E, Schrag TA, Melchinger AE (2013) Genetic diversity analysis of elite European maize (Zea mays L.) inbred lines using AFLP, SSR, and SNP markers reveals ascertainment bias for a subset of SNPs. Theor Appl Genet 126:133-141 Fu Y-B (2014) Genetic diversity analysis of highly incomplete SNP genotype data with imputations: an empirical assessment. G3: Genes|Genomes|Genetics doi: 10.1534/g3.114.010942 Fu Y-B, Cheng B, Peterson G (2014) Genetic diversity analysis of yellow mustard (Sinapis alba L.) germplasm based on genotyping by sequencing. Genet Resour Crop Evol 61:579-594 Fulling EH (1943) Plant life and the law of man. IV. Barberry, currant and gooseberry, and cedar control. Botanical Review 9:483-592 Ghazvini H, Hiebert C, Zegeye T, Liu S, Dilawari M, Tsilo T, Anderson J, Rouse M, Jin Y, Fetch T (2012) Inheritance of resistance to Ug99 stem rust in wheat cultivar Norin 40 and genetic mapping of Sr42. Theoretical and Applied Genetics 125:817-824 Gill BS, C. SH, Raupp WJ, Browder LE, Hatchett JH, Harvey TL, Moseman GJ, Waines JG (1985) Evaluation of Aegilops species for resistance to wheat powdery mildew, wheat leaf rust, hessian fly and greenbug. Plant Disease 69:314–316 125

Gold J, Harder D, Townley-Smith F, Aung T, Procunier J (1999) Development of a molecular marker for rust resistance genes Sr39 and Lr35 in wheat breeding lines. Electronic Journal of Biotechnology 2:1-6 Green GJ, Martens JW, Ribeiro O (1969) Epidemiology and specialization of wheat and oat stem rusts in Kenya in 1968. Phytopathology 60:309-314 Guo B, Beavis W (2011) In silico genotyping of the maize nested association mapping population. Mol Breeding 27:107-113 Guo B, Sleper DA, Beavis WD (2010) Nested association mapping for identification of functional markers. Genetics 186:373-383 Gupta PK, Kumar J, Mir RR, Kumar A (2010) Marker-assisted selection as a component of conventional plant breeding. Plant Breeding Reviews. John Wiley & Sons, Inc., pp 145-217 Haile JK, Röder MS (2013) Status of genetic research for resistance to Ug99 race of Puccinia graminis f. sp. tritici: A review of current research and implications. African Journal of Agricultural Research 8:6670-6680 Hamilton LM (1939) Stem rust in the spring wheat area in 1878. Minnesota History 20:156–164 Harder DE, Mathenge GR, Mwaura LK (1971) Physiologic specialization and epidemiology of wheat stem rust in East Africa. Phytopathology 62:166-171 Hare RA, McIntosh RA (1979) Genetic and cytogenetic studies of durable adult-plant resistances in Hope and related cultivars to wheat rusts. Zeitschrift für Pflanzenzüchtung 83:350-367 Hayden MJ, Kuchel H, Chalmers KJ (2004) Sequence tagged microsatellites for the Xgwm533 locus provide new diagnostic markers to select for the presence of stem rust resistance gene Sr2 in bread wheat (Triticum aestivum L.). Theoretical and Applied Genetics 109:1641-1647 Hayes HK, Ausemus ER, Stakman EC, Bailey CH, Wilson HK, Bamberg RH, Morkley MC, Crim RF, Levine MN (1936) Thatcher wheat. Station Bulletin - Minnesota Agriculture Experimental Station 325:1–36 Hayes HK, Stakman EC, Aamodt OS (1925) Inheritance in wheat of resistance to black stem rust. Phytopathology 15:371-387 Herrera-Foessel S, Singh R, Lillemo M, Huerta-Espino J, Bhavani S, Singh S, Lan C, Calvo-Salazar V, Lagudah E (2014) Lr67/Yr46 confers adult plant resistance to stem rust and powdery mildew in wheat. Theoretical and Applied Genetics 127:781-789 126

Heslot N, Rutkoski J, Poland J, Jannink J-L, Sorrells ME (2013) Impact of marker ascertainment bias on genomic selection accuracy and estimates of genetic diversity. PLoS ONE 8:e74612 Hiebert C, Fetch T, Zegeye T, Thomas J, Somers D, Humphreys DG, McCallum B, Cloutier S, Singh D, Knott D (2011) Genetics and mapping of seedling resistance to Ug99 stem rust in Canadian wheat cultivars ‘Peace’ and ‘AC Cadillac’. Theor Appl Genet 122:143-149 Hodson DP, Nazari K, Park RF, Hansen J, Lassen P, Arista J, Fetch T, Hovmøller M, Jin Y, Pretorius ZA, Sonder K (2011) Putting Ug99 on the map: an update on current and future monitoring. http://www.globalrust.org/db/attachments/about/309/1/Hodson.pdf [Accessed on 0121-2014] Jia J, Zhao S, Kong X, Li Y, Zhao G, He W, Appels R, Pfeifer M, Tao Y, Zhang X, Jing R, Zhang C, Ma Y, Gao L, Gao C, Spannagl M, Mayer KF, Li D, Pan S, Zheng F, Hu Q, Xia X, Li J, Liang Q, Chen J, Wicker T, Gou C, Kuang H, He G, Luo Y, Keller B, Xia Q, Lu P, Wang J, Zou H, Zhang R, Xu J, Gao J, Middleton C, Quan Z, Liu G, Wang J, Yang H, Liu X, He Z, Mao L, Wang J (2013) Aegilops tauschii draft genome sequence reveals a gene repertoire for wheat adaptation. Nature 496:91-95 Jin Y (2005) Races of Puccinia graminis identified in the United States during 2003. Plant Disease 89:1125-1127 Jin Y, Szabo LJ, Pretorius ZA, Singh RP, Ward R, Fetch T (2008) Detection of virulence to resistance gene Sr24 within race TTKS of Puccinia graminis f. sp. tritici. Plant Disease 92:923-926 Jin Y, Szabo LJ, Rouse MN, Fetch T, Pretorius ZA, Wanyera R, Njau P (2009a) Detection of virulence to resistance gene Sr36 within the TTKS race lineage of Puccinia graminis f. sp. tritici. Plant Disease 93:367–370 Jin Y, Szabo LJ, Rouse MN, G. FT, Pretorius ZA, Wanyera R, Njau P (2009b) Detection of virulence to resistance gene Sr36 within the TTKS race lineage of Puccinia graminis f. sp. tritici. Plant Disease 93:367-370 Johnston SJ, Sharp PJ, McIntosh RA (1998) Molecular markers for the Sr2 stem rust resistance gene. In: Slinkard AE (ed) Proceedings of the 9th International Wheat Genetics Symposium. University Extension Press, University of Saskatchewan, Saskatoon, Canada, pp 78-80 Jones SS, Dvořák J, Knott DR, Qualset CO (1991) Use of double-ditelosomic and normal chromosome 1D recombinant substitution lines to map Sr33 on chromosome arm 1DS in wheat. Genome 34:505-508 127

Joshi LM, Palmer LT (1973) Epidemiology of stem, leaf and stripe rusts of wheat in northern India. The Plant Disease Reporter 57:8–12 Jourjon M-F, Jasson S, Marcel J, Ngom B, Mangin B (2005) MCQTL: multi-allelic QTL mapping in multi-cross design. Bioinformatics 21:128-130 Kaur J, Bansal U, Khanna R, Saini R, Bariana H (2009) Molecular mapping of stem rust resistance in HD2009/WL711 recombinant inbred line population. International Journal of Plant Breeding 3:77-33 Kerber ER, Dyck PL (1973) Inheritance of stem rust resistance transferred from diploid wheat (Triticum monococum) to tetraploid and hexaploid wheat and chromosome location of the gene involved. Canadian Journal of Genetics and Cytology 15:397409 Kerber ER, Dyck PL (1990) Transfer to hexaploid wheat of linked genes for adult-plant leaf rust and seedling stem rust resistance from an amphiploid of Aegilops speltoides × Triticum monococcum. Genome 33:530-537 Kidwell K, Osborn T (1992) Simple plant DNA isolation procedures. In: Beckmann JS, Osborn TC (eds) Plant Genomes: Methods for Genetic and Physical Mapping. Springer Netherlands, pp 1-13 Kislev ME (1982) Stem rust of wheat 3300 years old found in Israel. Science 216:993994 Klindworth DL, Miller JD, Williams ND, Xu SS (2011) Resistance to recombinant stem rust race TPPKC in hard red spring wheat. Theoretical and Applied Genetics 123:603-613 Klindworth DL, Niu Z, Chao S, Friesen TL, Jin Y, Faris JD, Cai X, Xu SS (2012) Introgression and characterization of a goatgrass gene for a high level of resistance to Ug99 stem rust in tetraploid wheat. G3: Genes|Genomes|Genetics 2:665-673 Knott DR (1962) Inheritance of rust resistance. VIII. Additional studies on Kenya varieties of wheat. Crop Science 2:130-132 Knott DR (1968) The inheritance of resistance to stem rust races 56 and 15B-IL (Can.) in the wheat varieties Hope and H-44. Canadian Journal of Genetics and Cytology 10:311-320 Knott DR (1982) Multigenic inheritance of stem rust resistance in wheat. Crop Sci 22:393-399 Knott DR (1989) The Wheat Rusts — Breeding for Resistance. Theoretical and Applied Genetics 12:201 128

Knott DR (1990) Near-isogenic lines of wheat carrying genes for stem rust resistance. Crop Sci 30:901-905 Knott DR (2001) The relationship between seedling and field resistance to two races of stem rust in Thatcher wheat. Canadian Journal of Plant Science 81:415-418 Knott DR, Anderson RG (1956) The inheritance of rust resistance. I. The inheritance of stem rust resistance in ten varieties of common wheat. Canadian Journal of Agricultural Science 36:174-195 Kosambi DD (1943) The estimation of map distance from recombination values. Annals of Eugenics 12:172-175 Lagudah ES, McFadden H, Singh RP, Huerta-Espino J, Bariana HS, Spielmeyer W (2006) Molecular genetic characterization of the Lr34/Yr18 slow rusting resistance gene region in wheat. Theor Appl Genet 114:21-30 Letta T, Maccaferri M, Badebo A, Ammar K, Ricci A, Crossa J, Tuberosa R (2013) Searching for novel sources of field resistance to Ug99 and Ethiopian stem rust races in durum wheat via association mapping. Theoretical and Applied Genetics 126:1237-1256 Letta T, Olivera P, Maccaferri M, Jin Y, Ammar K, Badebo A, Salvi S, Noli E, Crossa J, Tuberosa R (2014) Association mapping reveals novel stem rust resistance loci in durum wheat at the seedling stage. Plant Gen 0:Li B, Kimmel M (2013) Factors influencing ascertainment bias of microsatellite allele sizes: impact on estimates of mutation rates. Genetics 195:563-572 Lindhout P (2002) The perspectives of polygenic resistance in breeding for durable disease resistance. Euphytica 124:217-226 Ling H-Q, Zhao S, Liu D, Wang J, Sun H, Zhang C, Fan H, Li D, Dong L, Tao Y, Gao C, Wu H, Li Y, Cui Y, Guo X, Zheng S, Wang B, Yu K, Liang Q, Yang W, Lou X, Chen J, Feng M, Jian J, Zhang X, Luo G, Jiang Y, Liu J, Wang Z, Sha Y, Zhang B, Wu H, Tang D, Shen Q, Xue P, Zou S, Wang X, Liu X, Wang F, Yang Y, An X, Dong Z, Zhang K, Zhang X, Luo M-C, Dvorak J, Tong Y, Wang J, Yang H, Li Z, Wang D, Zhang A, Wang J (2013) Draft genome of the wheat A-genome progenitor Triticum urartu. Nature 496:87-90 Liu H, Bayer M, Druka A, Russell J, Hackett C, Poland J, Ramsay L, Hedley P, Waugh R (2014) An evaluation of genotyping by sequencing (GBS) to map the Breviaristatum-e (ari-e) locus in cultivated barley. BMC Genomics 15:104 Liu K, Muse SV (2005) PowerMarker: an integrated analysis environment for genetic marker analysis. Bioinformatics 21:2128-2129 129

Liu S, Yu L-X, Singh R, Jin Y, Sorrells M, Anderson J (2010) Diagnostic and codominant PCR markers for wheat stem rust resistance genes Sr25 and Sr26. Theoretical and Applied Genetics 120:691-697 Liu W, Rouse M, Friebe B, Jin Y, Gill B, Pumphrey M (2011) Discovery and molecular mapping of a new gene conferring resistance to stem rust, Sr53, derived from Aegilops geniculata and characterization of spontaneous translocation stocks with reduced alien chromatin. Chromosome Res 19:669-682 Lorieux M (2012) MapDisto: fast and efficient computation of genetic linkage maps. Mol Breeding 30:1231-1235 Lu F, Lipka AE, Glaubitz J, Elshire R, Cherney JH, Casler MD, Buckler ES, Costich DE (2013) Switchgrass genomic diversity, ploidy, and evolution: novel insights from a network-based SNP discovery protocol. PLoS Genet 9:e1003215 Lukaszewski AJ, Curtis CA (1993) Physical distribution of recombination in B-genome chromosomes of tetraploid wheat. Theoretical and Applied Genetics 86:121-127 Macharia GK (2013) Molecular diversity, linkage disequilibrium and genetic mapping in East Africa wheat. Department of Agronomy. University of Minnesota, St Paul, MN, USA Mago R, Bariana HS, Dundas IS, Spielmeyer W, Lawrence GJ, Pryor AJ, Ellis JG (2005) Development of PCR markers for the selection of wheat stem rust resistance genes Sr24 and Sr26 in diverse wheat germplasm. Theoretical and Applied Genetics 111:496-504 Mago R, Brown-Guedira G, Dreisigacker S, Breen J, Jin Y, Singh R, Appels R, Lagudah ES, Ellis J, Spielmeyer W (2011a) An accurate DNA marker assay for stem rust resistance gene Sr2 in wheat. Theoretical and Applied Genetics 122:735-744 Mago R, Lawrence G, Ellis J (2011b) The application of DNA marker and doubledhaploid technology for stacking multiple stem rust resistance genes in wheat. Mol Breeding 27:329-335 Mago R, Zhang P, Bariana HS, Verlin DC, Bansal UK, Ellis JG, Dundas IS (2009) Development of wheat lines carrying stem rust resistance gene Sr39 with reduced Aegilops speltoides chromatin and simple PCR markers for marker-assisted selection. Theoretical and Applied Genetics 119:1441-1450 MAS Wheat (2014) http://maswheat.ucdavis.edu/protocols/StemRust/index.htm. Accessed on 05-24-2014. UC-Davis Mayfield AH (1985) Efficacies of fungicides for control of stem rust of wheat. Australian Journal of Experimental Agriculture 25:440–443 130

McAlpine D (1906) The rusts of Australia. Their structure, nature, and classification. Department of Agriculture, Melbourne, Australia McDonald J (1931) The existence of physiologic forms of wheat stem rust in Africa. Transactions of the British Mycological Society 15:235 McFadden ES (1930) A successful transfer of emmer characters to vulgare wheat. Journal of the American Society of Agronomy 22:1020-1034 McFadden ES (1939) Brown necrosis, a discoloration associated with rust infection in certain rust resistant wheats. Journal of Agricultural Research 58:805–819 McIntosh RA, Dubcovsky J, Rojers WJ, Morris C, Sommers DJ, Appels R, Xia XC (2011) Catalogue of gene symbols for wheat: 2011 supplement. Annual Wheat Newsletter 57:303–321 McIntosh RA, Gyarfas J (1971) Triticum timopheevi as a source of resistance to wheat stem rust. Zeitschrift für Pflanzenzüchtung 66:240-248 McIntosh RA, Wellings CR, Park RF (1995) Wheat rusts: an atlas of resistance genes. CSIRO Publications, Victoria, Australia McIntosh RA, Yamazaki Y, Dubcovsky J, Rogers J, Morris C, Somers DJ, Appels R, Devos KM (2003) Catalogue of gene symbols for wheat. 11th International Wheat Genetics Symposium, Brisbane, Australia McMullen MD, Kresovich S, Villeda HS, Bradbury P, Li H, Sun Q, Flint-Garcia S, Thornsberry J, Acharya C, Bottoms C, Brown P, Browne C, Eller M, Guill K, Harjes C, Kroon D, Lepak N, Mitchell SE, Peterson B, Pressoir G, Romero S, Rosas MO, Salvo S, Yates H, Hanson M, Jones E, Smith S, Glaubitz JC, Goodman M, Ware D, Holland JB, Buckler ES (2009) Genetic properties of the maize nested association mapping population. Science 325:737-740 Moragues M, Comadran J, Waugh R, Milne I, Flavell AJ, Russell JR (2010) Effects of ascertainment bias and marker number on estimations of barley diversity from highthroughput SNP genotype data. Theor Appl Genet 120:1525-1534 Moser G, Khatkar MS, Raadsma HW (2009) Imputation of missing genotypes in high density SNP data. Proceedings of 18th Conference Of The Association For The Advancement Of Animal Breeding And Genetics, Barrosa Valley, Australia, pp 612615 Mundt CC (2014) Durable resistance: A key to sustainable management of pathogens and pests. Infection, Genetics and Evolution DOI: 10.1016/j.meegid.2014.01.011

131

Nazareno NRX, Roelfs AP (1981) Adult plant resistance of Thatcher wheat to stem rust. Phytopathology 71:181-185 Negeri A, Coles N, Holland J, Balint-Kurti P (2011) Mapping QTL controlling southern leaf blight resistance by joint analysis of three related recombinant inbred line populations. Crop Science 51:1571-1579 Nelson JC, Deynze AEV, Sorrells ME, Autrique E, Lu YH, Merlino M, Atkinson M, Leroy P (1995) Molecular mapping of wheat homoeologous group 2. Genome 38:516-524 Njau PN, Jin Y, Huerta-Espino J, Keller B, Singh RP (2010) Identification and evaluation of sources of resistance to stem rust race Ug99 in wheat. Plant Disease 94:413-419 Njau PN, Wanyera R, Macharia GK, Macharia J, Singh R, Keller B (2009) Resistance in Kenyan bread wheat to recent eastern African isolate of stem rust, Puccinia graminis f. sp. tritici, Ug99. Journal of Plant Breeding and Crop Science 1:22-27 Olson EL, Brown-Guedira G, Marshall D, Stack E, Bowden RL, Jin Y, Rouse M, Pumphrey MO (2010a) Development of wheat lines having a small introgressed segment carrying stem rust resistance gene Sr22. Crop Science 50:1823-1830 Olson EL, Brown-Guedira G, Marshall DS, Jin Y, Mergoum M, Lowe I, Dubcovsky J (2010b) Genotyping of U.S. wheat germplasm for presence of stem rust resistance genes Sr24, Sr36 and Sr1RSAmigo. Crop Science 50:668-675 Parlevliet JE (1976) Partial resistance of barley to leaf rust, Puccinia hordei. III. The inheritance of the host plant effect on latent period in four cultivars. Euphytica 25:241-248 Paull JG, Pallotta MA, Langridge P, The TT (1994) RFLP markers associated with Sr22 and recombination between chromosome 7A of bread wheat and the diploid species Triticum boeoticum. Theoretical and Applied Genetics 89:1039-1045 Periyannan S, Bansal U, Bariana H, Deal K, Luo M-C, Dvorak J, Lagudah E (2014) Identification of a robust molecular marker for the detection of the stem rust resistance gene Sr45 in common wheat. Theoretical and Applied Genetics 127:947955 Periyannan S, Moore J, Ayliffe M, Bansal U, Wang X, Huang L, Deal K, Luo M, Kong X, Bariana H, Mago R, McIntosh R, Dodds P, Dvorak J, Lagudah E (2013) The gene Sr33, an ortholog of barley Mla genes, encodes resistance to wheat stem rust race Ug99. Science 341:786-788 Persoon CH ( 1794) Neuer versuch einer systematischen eintheilung der schwämme. Roemer's Neues Magazin für die Botanik 1:63–128 132

Peterson RF, Campbell AB, Hannah AE (1948) A diagramatic scale for estimating rust intensity of leaves and stem of cereals. Canadian Journal of Research 26c:496-500 Plessers AG (1954) The genetics of stem and leaf rust reactions and other characters in Lee wheat with Chinese monosomic testers. Dissertation Abstracts 15 Poland JA, Brown PJ, Sorrells ME, Jannink J-L (2012a) Development of high-density genetic maps for barley and wheat using a aovel two-enzyme genotyping-bysequencing approach. PLoS ONE 7:e32253 Poland JA, Brown PJ, Sorrells ME, Jannink J-L (2012b) Development of high-density genetic maps for barley and wheat using a novel two-enzyme gnotyping-bysequencing approach. PLoS ONE 7:e32253 Pozniak CJ, Reimer S, Fetch T, Clarke JM, Clarke FR, Somers DJ, Knox RE, Singh AK (2008) Association mapping of Ug99 resistance in diverse durum wheat population. In: Rudi A, Russell E, Peter L, Michael M, Lynne M, Peter S (eds) 11th International Wheat Genetics Symposium, Brisbane, Australia, pp 809–811 Pretorius ZA, Singh RP, Wagoire WW, Payne TS (2000) Detection of virulence to wheat stem rust resistance gene Sr31 in Puccinia graminis f. sp. tritici in Uganda. Plant Disease 84 Pretorius ZA, Szabo LJ, Boshoff WHP, Herselman L, Visser B (2012) First report of a new TTKSF race of wheat stem rust (Puccinia graminis f. sp. tritici) in South Africa and Zimbabwe. Plant Disease 96:590-590 Prins R, Pretorius ZA, Bender CM, Lehmensiek A (2011a) QTL mapping of stripe, leaf and stem rust resistance genes in a 'Kariega' × 'Avocet S' doubled haploid wheat population. Mol Breeding 27:259–270 Prins R, Pretorius ZA, Bender CM, Lehmensiek A (2011b) QTL mapping of stripe, leaf and stem rust resistance genes in a Kariega × Avocet S doubled haploid wheat population. Mol Breeding 27:259-270 Qi LL, Pumphrey MO, Friebe B, Zhang P, Qian C, Bowden RL, Rouse MN, Jin Y, Gill BS (2011) A novel Robertsonian translocation event leads to transfer of a stem rust resistance gene (Sr52) effective against race Ug99 from Dasypyrum villosum into bread wheat. Theor Appl Genet 123:159-167 Rebai A, Goffinet B (1993) Power of tests for QTL detection using replicated progenies derived from a diallel cross. Theoretical and Applied Genetics 86:1014-1022 Remington DL, Ungerer MC, Purugganan MD (2001) Map-based cloning of quantitative trait loci: progress and prospects. Genetics Research 78:213-218 133

Risk JM, Selter LL, Chauhan H, Krattinger SG, Kumlehn J, Hensel G, Viccars LA, Richardson TM, Buesing G, Troller A, Lagudah ES, Keller B (2013) The wheat Lr34 gene provides resistance against multiple fungal pathogens in barley. Plant Biotechnol J 11:847-854 Ritland K (1996) Estimators for pairwise relatedness and individual inbreeding coefficients. Genetical Research 67:175-185 Roelfs AP (1977) Foliar fungal diseases of wheat in the People’s Republic of China. The Plant Disease Reporter 61:836–841 Roelfs AP (1985a) Epidemiology in North America. In 'The Cereal Rusts'. Vol. II. Academic Press, Orlando, Florida, USA Roelfs AP (1985b) Wheat and rye stem rust. In: Roelfs AP, Bushnell WR (eds) The Cereal Rusts. Academic Press, Orlando, FL, USA, pp 3–37 Roelfs AP, Martens JW (1988) An international system of nomenclature for Puccinia graminis f. sp. tritici. Phytopathology 78:526-533 Roelfs AP, McVey DV (1979) Low infection types produced by Puccinia graminis f.sp. tritici and wheat lines with designated genes for resistance. Phytopathology 69:722730 Roelfs AP, Singh RP, Saari EE (1992) Rust diseases of wheat: Concepts and methods of disease management. CIMMYT, Mexico Rouse M, Nava I, Chao S, Anderson J, Jin Y (2012) Identification of markers linked to the race Ug99 effective stem rust resistance gene Sr28 in wheat (Triticum aestivum L.). Theoretical and Applied Genetics 125:877-885 Rouse MN, Jin Y (2011) Stem rust resistance in A-genome diploid relatives of wheat. Plant Disease 95:941-944 Rouse MN, Talbert LE, Singh D, Sherman JD (2014) Complementary epistasis involving Sr12 explains adult plant resistance to stem rust in Thatcher wheat (Triticum aestivum L.). Theor Appl Genet 127:1549-1559 Rutkoski JE, Poland J, Jannink JL, Sorrells ME (2013) Imputation of unordered markers and the impact on genomic selection accuracy. G3 (Bethesda, Md) 3:427-439 Saintenac C, Jiang D, Wang S, Akhunov E (2013a) Sequence-based mapping of the polyploid wheat genome. G3: Genes|Genomes|Genetics 3:1105-1114

134

Saintenac C, Zhang W, Salcedo A, Rouse MN, Trick HN, Akhunov E, Dubcovsky J (2013b) Identification of wheat gene Sr35 that confers resistance to Ug99 stem rust race group. Science 341:783-786 Salvi S, Tuberosa R (2005) To clone or not to clone plant QTLs: present and future challenges. Trends in Plant Science 10:297-304 Sandhu D, Gill KS (2002) Gene-containing regions of wheat and the other grass genomes. Plant Physiology 128:803–811 Sawhney RN, Singh D, Joshi BC (1981) Monosomic analysis of genes for resistance against stem rust races in bread wheat. Theoretical and Applied Genetics 59:313-316 Seah S, Spielmeyer W, Jahier J, Sivasithamparam K, Lagudah ES (2000) Resistance gene analogs within an introgressed chromosomal segment derived from Triticum ventricosum that confers resistance to nematode and rust pathogens in wheat. Molecular Plant-Microbe Interactions 13:334-341 Sears ER (1954) The aneuploids of common wheat. University of Missouri Research Bulletin 572 Sears ER, Loegering WQ, Rodenhiser HA (1957) Identification of chromosomes carrying genes for stem rust resistance in four varieties of wheat. Agronomy Journal 49:208212 Sharp PJ, Johnston S, Brown G, McIntosh RA, Pallotta M, Carter M, Bariana HS, Khatkar S, Lagudah ES, Singh RP, Khairallah M, Potter R, Jones MGK (2001) Validation of molecular markers for wheat breeding. Australian Journal of Agricultural Research 52:1357-1366 Sidhu D, Gill KS (2004) Distribution of genes and recombination in wheat and other eukaryotes. Plant Cell, Tissue and Organ Culture 4653PB:1-14 Simons K, Abate Z, Chao S, Zhang W, Rouse M, Jin Y, Elias E, Dubcovsky J (2011) Genetic mapping of stem rust resistance gene Sr13 in tetraploid wheat (Triticum turgidum ssp. durum L.). Theoretical and Applied Genetics 122:649-658 Singh A, Knox RE, DePauw RM, Singh AK, Cuthbert RD, Campbell HL, Singh D, Bhavani S, Fetch T, Clarke F (2013a) Identification and mapping in spring wheat of genetic factors controlling stem rust resistance and the study of their epistatic interactions across multiple environments. Theor Appl Genet 126:1951-1964 Singh A, Pandey MP, Singh AK, Knox RE, Ammar K, Clarke JM, Clarke FR, Singh RP, Pozniak CJ, DePauw RM, McCallum BD, Cuthbert RD, Randhawa HS, Fetch TG, Jr. (2013b) Identification and mapping of leaf, stem and stripe rust resistance 135

quantitative trait loci and their interactions in durum wheat. Mol Breeding 31:405418 Singh RP (2012) Pros and cons of utilizing major, race-specific resistance genes versus partial resistance in breeding rust resistant wheat. International Maize and Wheat Improvement Center (CIMMYT), Mexico Singh RP, Herrera-Foessel SA, Huerta-Espino J, Lan CX, Basnet BR, Bhavani S, Lagudah ES (2013c) Pleiotropic gene Lr46/Yr29/Pm39/Ltn2 confers slow rusting, adult plant resistance to wheat stem rust fungus. Borlaug Global Rust Initiative 2013 Technical Workshop, New Delhi, India, p 17 Singh RP, Hodson DP, Huerta-Espino J, Jin Y, Bhavani S, Njau P, Herrera-Foessel S, Singh PK, Singh S, Govindan V (2011) The emergence of Ug99 races of the stem rust fungus is a threat to world wheat production. Annual Review of Phytopathology 49:465-481 Singh RP, Hodson DP, Huerta-Espino J, Jin Y, Njau P, Wanyera R, Herrera-Foessel SA, Ward RW (2008a) Will stem rust destroy the world's wheat crop? In: Donald LS (ed) Advances in Agronomy. Academic Press, pp 271-309 Singh RP, Hodson DP, Huerta-Espino J, Jin Y, Njau P, Wanyera R, Herrera-Foessel SA, Ward RW (2008b) Will stem rust destroy the world's wheat crop? Advances in Agronomy 98:272-309 Singh RP, Hodson DP, Jin Y, Huerta-Espino J, Kinyua MG, Wanyera R, Njau P, Ward RW (2006) Current status, likely migration and strategies to mitigate the threat to wheat production from race Ug99 (TTKS) of stem rust pathogen. CAB Reviews: Perspectives in Agriculture, Veterinary Science, Nutrition and Natural Resources 1 Singh RP, Huerta-Espino J, Rajaram S (2000) Achieving near-immunity to leaf and stripe rusts in wheat by combining slow rusting resistance genes. Acta Phytopathologica et Entomologica Hungarica 35:133-139 Singh RP, Huerta EJ, Manilal HM (2005) Genetics and breeding for durable resistance to leaf and stripe rusts in wheat. Turkish Journal of Agriculture and Forestry 29:121127 Singh S, Singh R, Bhavani S, Huerta-Espino J, Eugenio L-V (2013d) QTL mapping of slow-rusting, adult plant resistance to race Ug99 of stem rust fungus in PBW343/Muu RIL population. Theor Appl Genet 126:1367-1375 Spielmeyer W, Sharp PJ, Lagudah ES (2003) Identification and validation of markers linked to broad-spectrum stem rust resistance gene Sr2 in wheat (Triticum aestivum L.). Crop Sci 43:333-336 136

Stacklies W, Redestig H, Scholz M, Walther D, Selbig J (2007) pcaMethods—a bioconductor package providing PCA methods for incomplete data. Bioinformatics 23:1164-1167 Stakman EC, Harrar JG (1957) Principles of plant pathology. Ronald Press, New York Stakman EC, Stewart DM, Loegering WQ (1962) Identification of physiologic races of Puccinia graminis var. tritici. US Dep Agric Agric Res Serv E 6/7 Steffenson BJ, Olivera P, Roy JK, Jin Y, Smith KP, Muehlbauer GJ (2007) A walk on the wild side: mining wild wheat and barley collections for rust resistance genes. Australian Journal of Agricultural Research 58:532-544 Stubbs RW, Prescott JM, Saari EE, Dubin HJ (1986) Cereal disease methodology manual, CIMMYT, Mexico Stuthman DD, Leonard KJ, Miller‐Garvin J (2007) Breeding crops for durable resistance to disease. In: Donald LS (ed) Advances in Agronomy. Academic Press, pp 319-367 Suenaga K, Singh RP, Huerta-Espino J, William HM (2003) Microsatellite markers for genes Lr34/Yr18 and other quantitative trait Loci for leaf rust and stripe rust resistance in bread wheat. Phytopathology 93:881-890 Targioni TG (1767) True nature, causes and sad effects of the rust, the bunt, the smut, and other maladies of wheat and of oats in the field. American Phytopathological Society, St Paul, Minnesota, USA Theophrastus Enquiry into plants II. Harvard University Press Cambridge, Massachussetts Tsilo TJ, Jin Y, Anderson JA (2007) Microsatellite markers linked to stem rust resistance allele Sr9a in wheat. Crop Science 47:2013-2020 Tsilo TJ, Kolmer JA, Anderson JA (2014) Molecular mapping and improvement of leaf rust resistance in wheat breeding lines. Phytopathology 104:865-870 Tulasne LR (1853) Note sur la germination des spores des uredinees. comptes rendus hebdomadaires des séances de l'Académie des Sciences 36:1093-1095 [Translation in Annals and Magazine of Natural History 2, 221-223 (1853)]. Tulasne LR, Tulasne C (1847) Mémoire sur les ustilaginees comparees aux Wang J, Li H, Zhang L, Meng L (2012a) Users’ manual of QTL IciMapping Version 3.3. The Quantitative Genetics Group, Institute of Crop Science, Chinese Academy of Agricultural Sciences (CAAS), China, and Genetic Resources Program, International Maize and Wheat Improvement Center (CIMMYT), Mexico 137

Wang S, Basten CJ, Zeng Z-B (2012b) Windows QTL Cartographer 2.5. Department of Statistics, North Carolina State University, Raleigh, NC Wang S, Wong D, Forrest K, Allen A, Chao S, Huang BE, Maccaferri M, Salvi S, Milner SG, Cattivelli L, Mastrangelo AM, Whan A, Stephen S, Barker G, Wieseke R, Plieske J, International Wheat Genome Sequencing C, Lillemo M, Mather D, Appels R, Dolferus R, Brown-Guedira G, Korol A, Akhunova AR, Feuillet C, Salse J, Morgante M, Pozniak C, Luo M-C, Dvorak J, Morell M, Dubcovsky J, Ganal M, Tuberosa R, Lawley C, Mikoulitch I, Cavanagh C, Edwards KJ, Hayden M, Akhunov E (2014) Characterization of polyploid wheat genomic diversity using a high-density 90 000 single nucleotide polymorphism array. Plant Biotechnology Journal:1-10 Wanyera R, Kinyua MG, Jin Y, Singh RP (2006) The spread of stem rust caused by Puccinia graminis f. sp. tritici, with virulence on Sr31 in wheat in Eastern Africa. Plant Disease 90:113-113 Wanyera R, Macharia JK, Kilonzo SM, Kamundia JW (2009) Foliar fungicides to control wheat stem rust, race TTKS (Ug99), in Kenya. Plant Disease 93:929-932 Waterhouse WL (1929) Australian rust studies. I. Proceedings of the Linnean Society of New South Wales 54:615–680 Watson IA (1981) Wheat and its rust parasites in Australia. In: Evans LT, Peacock WJ (eds) Wheat Science – Today and Tomorrow. Cambridge University Press, London, UK, pp 129–147 Werner JE, Endo TR, Gill BS (1992) Toward a cytogenetically based physical map of the wheat genome. Proceedings of the National Academy of Sciences of the United States of America 89:11307-11311 Wheat MAS (2014) http://maswheat.ucdavis.edu/protocols/StemRust/index.htm. Accessed on 05-24-2014. UC-Davis William M, Singh RP, Huerta-Espino J, Islas SO, Hoisington D (2003) Molecular marker mapping of leaf rust resistance gene Lr46 and its association with stripe rust resistance gene Yr29 in wheat. Phytopathology 93:153-159 Williams L, Ma X, Boyko A, Bustamante C, Oleksiak M (2010) SNP identification, verification, and utility for population genetics in a non-model genus. BMC Genetics 11:32 Yu J, Holland JB, McMullen MD, Buckler ES (2008) Genetic design and statistical power of nested association mapping in maize. Genetics 178:539-551

138

Yu L-X, Barbier H, Rouse M, Singh S, Singh R, Bhavani S, Huerta-Espino J, Sorrells M (2014) A consensus map for Ug99 stem rust resistance loci in wheat. Theor Appl Genet 127:1561-1581 Yu L-X, Lorenz A, Rutkoski J, Singh R, Bhavani S, Huerta-Espino J, Sorrells M (2011) Association mapping and gene–gene interaction for stem rust resistance in CIMMYT spring wheat germplasm. Theor Appl Genet 123:1257-1268 Yu L-X, Morgounov A, Wanyera R, Keser M, Singh S, Sorrells M (2012) Identification of Ug99 stem rust resistance loci in winter wheat germplasm using genome-wide association analysis. Theor Appl Genet 125:749-758 Zadoks JC (1963) Epidemiology of wheat rusts in Europe. FAO Plant Protection Bulletin 13:97–108 Zadoks JC, Chang TT, Konzak CF (1974) A decimal code for the growth stages of cereals. Weed Research 14:415-421 Zeller FJ (1973) 1B/1R wheat-rye chromosome substitutions and translocations. In: Sears ER, Sears LMS (eds) Proceedings of the Fourth International Wheat Genetics Symposium. Agricultural Experiment Station, University of Missouri, Columbia, Missouri, USA, pp 209-221 Zhang D, Bowden RL, Yu J, Carver BF, Bai G (2014) Association analysis of stem rust resistance in U.S. winter wheat. PLoS ONE 9:e103747 Zhensheng K, Jie Z, Dejun H, Hongchang Z, Xiaojie W, Chenfang W, Qingmei H, Jun G, Lili H (2010) Status of wheat rust research and control in China. BGRI 2010 Technical Workshop 30-31 May, St. Petersburg, Russia

139

APPENDIX

Appendix I: QTL detected via CIM in the 9K (Table A), non-imputed GBS (Table B), GBS40 (Table C), and GBS75 (Table D) datasets. A. Environment

Chromosome

Ken12

LODa

R2b

Addc

Marker

Pos (cM)

2B 2B 4A 6D

26 93 9 2

24.7 122.5 9.2 2.4

15.7 3.1 3.4 2.9

32.6 6.0 6.6 5.1

-6.2 2.6 -2.8 2.4

Ken13

2B 4B 5B

17 9 3

14.2 25.8 0.4

4.8 2.6 4.0

23.7 11.0 18.5

-5.0 -3.3 5.3

Eth13

2B 3A

26 22

24.7 60.3

17.0 2.8

57.6 6.1

-11.2 -3.7

StP13

2B 4B 4B 4B

20 9 34 77

21.4 24.8 49.0 62.1

3.8 2.6 4.7 3.0

9.8 5.7 10.4 12.1

-2.6 2.1 -3.0 -3.0

Marker

Pos (cM)

B. Environment

Chromosome

LODa

R2b

Addc

Ken12

2B 2B 7A

23 104 6

15.0 93.9 0.8

4.1 15.4 4.1

7.1 32.1 6.7

-2.9 6.2 -2.8

Ken13

1A 2B 2D

57 107 7

92.9 95.7 32.2

3.4 5.2 3.3

14.3 29.6 14.7

3.8 5.6 -3.8

Eth13

2A 2B

11 105

30.6 94.5

2.8 13.8

7.4 49.6

3.9 10.4

140

Environment

Chromosome

Marker

Pos (cM)

LODa

R2b

Addc

Eth13

4B

17

33.6

2.7

5.7

3.6

StP13

2B 4A

115 40

107.1 21.8

3.8 2.6

8.8 6.0

2.5 2.1

Marker

Pos (cM)

C. Environment

Chromosome

LODa

R2b

Addc

Ken12

2B 2B 7A

17 104 5

58.9 307.1 24.4

5.0 17.2 4.4

8.3 36.1 7.2

-3.2 6.6 -2.9

Ken13

1A 2B 2B 2D

57 103 113 9

312.1 295.2 331.6 97.8

3.4 2.5 2.5 3.3

14.4 11.2 9.9 13.6

3.9 3.7 4.0 -3.7

Eth13

2B

105

308.7

14.4

52.6

10.7

StP13

2B 4A 4B

107 37 50

321.0 60.4 208.1

3.6 2.6 2.9

9.2 5.9 6.6

2.5 2.1 2.2

Marker

Pos (cM)

D. Environment

Chromosome

LODa

R2b

Addc

Ken12

2B 2B 7B

35 122 9

696.2 2237.3 167.2

2.8 16.3 3.6

4.6 34.1 6.1

-2.3 6.4 -2.7

Ken13

1A 2B 3A

56 121 18

1004.8 2228.4 377.1

3.3 2.5 3.2

14.4 11.8 13.5

3.9 3.8 -3.6

Eth13

2B

123

2238.2

15.2

53.5

10.8

StP13

2B 4B 5B

128 16 24

2328.3 172.3 567.3

3.7 2.8 3.0

9.4 6.2 6.9

2.6 2.1 2.3

141

a

LOD values are the peak logarithm of odds score for the given QTL

b

Value indicates the phenotypic variation explained by the QTL

c

Value indicates the estimated additive effect of the QTL; negative value means that the allele was contributed by RB07

142

Appendix II: Quantitative trait loci (QTL) detected for stem rust adult plant resistance by joint mapping of 10 RIL populations in four environments. Env Chra

QTLb

Left SNP

Right SNP

Posc

LODd

PVEe (%)

Adaf

Faharif

Gemf

Kuduf

Kulunguf

Ngirif

Pakaf

Pasaf

Popof

Romanyf

Ken13 2A

QSr.umn-2A.1

SNP126

SNP652

159.1

3.3

2.1

4.4

-0.7

-3.2

1.9

3.3

-3.4

-0.2

-1.5

-0.4

0.3

2B

QSr.umn-2B.1

SNP970

SNP358

444.8

4.2

2.6

-3.0

1.8

-0.6

2.1

2.2

-2.8

-0.2

-0.9

0.9

-0.4

3B

QSr.umn-3B.1

SNP666

SNP243

295.0

3.2

2.0

0.4

-0.9

-4.5

2.0

6.0

-2.8

-2.1

1.0

0.9

-0.2

3B

QSr.umn-3B.2

SNP4

SNP813

723.8

2.9

1.9

2.5

-1.3

2.9

-1.8

-4.7

0.4

1.8

1.0

1.8

-2.0

3B

QSr.umn-3B.3

SNP350

SNP353

787.2

3.7

2.3

-10.4

4.2

-0.8

-4.8

0.0

5.1

3.0

0.8

0.2

-0.1

4A

QSr.umn-4A.1

SNP371

SNP375

164.3

2.5

1.5

6.8

-1.2

-0.9

-6.5

-2.8

4.1

2.0

-3.0

1.5

0.8

4A

QSr.umn-4A.2

SNP973

SNP522

644.9

6.6

4.1

5.5

1.2

-2.8

1.9

-7.8

-2.7

-2.3

2.3

3.2

1.0

4B

QSr.umn-4B.1

SNP159

SNP435

401.3

2.8

0.6

-2.9

0.2

-0.8

0.0

0.2

0.8

0.8

1.0

2.7

-2.3

5B

QSr.umn-5B.1

SNP455

SNP150

607.2

2.5

1.6

-1.0

0.9

1.9

-0.6

1.3

-2.3

1.5

0.1

0.8

-3.1

6A

QSr.umn-6A

SNP935

SNP425

210.3

4.2

2.6

1.3

-26.7

1.2

4.7

3.5

0.0

3.8

2.0

1.0

6.9

7A

QSr.umn-7A

SNP191

SNP181

97.5

4.1

2.5

2.0

-1.9

-2.1

10.0

0.0

-5.2

-0.1

-1.8

-0.5

1.5

4B

QSr.umn-4B.2

SNP177

SNP203

358.8

7.3

4.0

-0.1

0.1

6.6

-0.8

-0.4

-0.4

0.4

-5.5

-0.2

0.4

7B

QSr.umn-7B

SNP268

SNP337

43.0

7.6

4.1

-1.0

0.7

2.2

2.0

0.6

2.2

-0.3

0.8

0.5

-8.1

2A

QSr.umn-2A.2

SNP537

SNP494

275.8

3.6

2.2

-5.4

7.6

-0.9

-13.5

4.7

5.3

8.1

4.8

3.2

1.5

2B

QSr.umn-2B.2

SNP294

SNP970

439.9

6.1

3.8

0.0

-2.7

-0.4

2.8

6.4

-4.8

-2.2

-1.2

0.9

-0.9

2D

QSr.umn-2D

SNP116

SNP687

185.1

2.9

1.8

2.4

-0.4

-1.3

0.1

-5.5

-3.9

1.5

2.7

-1.9

4.5

SA12

StP12

143

Env Chr

a

QTL

b

Left SNP

Right SNP

Pos

c

d

PVEe (%)

LOD

Adaf

Faharif

Gemf

Kuduf

Kulunguf

Ngirif

Pakaf

Pasaf

Popof

Romanyf

StP12 3B

QSr.umn-3B.4

SNP89

SNP395

89.4

2.7

1.4

-6

-1.9

10.7

-3.5

17.3

0.0

-5.2

-9.1

1.6

-4.1

3B

QSr.umn-3B.5

SNP154

SNP93

642.7

5.8

3.6

-0.7

-0.3

-2.2

-0.1

4.6

4.4

0.7

2.8

-11.5

0.6

3B

QSr.umn-3B.6

SNP790

SNP924

684.6

6.2

3.8

8.2

2.9

3.7

-1.6

-4.7

-5.2

-0.7

1.1

-0.8

-2.1

3B

QSr.umn-3B.7

SNP320

SNP359

811.0

8.3

5.0

-4.9

4.5

-26.2

-11.2

-0.6

6.8

3.9

0.1

19.2

2.5

4A

QSr.umn-4A.1

SNP371

SNP375

164.3

2.5

1.5

3.4

-4.7

-4.6

-4.3

0.3

8.0

2.3

1.4

-3.8

3.4

4B

QSr.umn-4B.3

SNP205

SNP210

301.4

4.3

2.6

3.0

5.3

-9.3

-1.6

-13.4

13.4

3.2

-0.3

1.5

-3.3

5B

QSr.umn-5B.2

SNP142

SNP784

624.2

2.8

1.8

-1.9

-0.8

4.7

-2.8

1.7

-1.4

-2.6

-1.7

3.0

1.7

5D

QSr.umn-5D.1

SNP714

SNP481

138.0

3.0

1.9

-4.3

0.6

0.7

-0.4

0.0

2.7

2.8

-3.2

-2.0

2.0

StP13 2A

QSr.umn-2A.1

SNP126

SNP652

159.1

2.8

1.3

1.3

-2.6

-3.0

2.0

2.2

-3.0

2.5

0.0

1.2

-1.7

2B

QSr.umn-2B.3

SNP840

SNP618

325.7

3.0

2.2

5.3

3.2

0.8

-0.8

2.8

-7.2

-4.1

0.9

0.5

-2.7

2B

QSr.umn-2B.2

SNP294

SNP970

434.9

3.2

2.3

3.5

-4.9

1.7

2.7

2.7

-1.4

-2.3

0.3

-1.1

-3.0

3B

QSr.umn-3B.4

SNP89

SNP395

89.4

3.2

2.3

-5.5

-5.4

15.3

-11.3

11.8

0.0

-8.7

-10.7

0.2

-10.4

3B

QSr.umn-3B.8

SNP885

SNP790

684.5

3.6

2.6

4.5

4.5

1.2

-2.7

-3.8

-0.3

0.7

1.7

-1.1

-4.1

3B

QSr.umn-3B.7

SNP320

SNP359

811.0

3.5

2.6

1.3

7.3

-13.4

-23.7

9.2

-0.9

4.6

1.3

3.9

3.7

4B

QSr.umn-4B.1

SNP159

SNP435

396.3

2.6

1.7

0.5

-2.1

-4.2

-0.2

0.7

2.2

-1.9

3.0

-1.1

3.8

5B

QSr.umn-5B.3

SNP141

SNP765

575.9

2.9

2.1

-1.2

2.9

2.2

6.2

-1.0

-7.0

-1.0

-1.8

1.0

-0.3

5D

QSr.umn-5D.2

SNP474

SNP977

101.4

2.6

1.9

-2.1

1.6

-3.1

-0.1

3.0

0.2

-1.7

-1.7

1.5

0.9

a

Chromosome location of the QTL

b

QTL were named according to McIntosh et al. (2003) 144

c

Position (centiMorgan) of the detected QTL peak in Chromosome ‘Chrom’

d

Logarithm of odds scores for the QTL detected at position ‘Pos’, based on joint mapping

e

Percentage of phenotypic variation explained by the observed QTL, based on joint mapping

f

Estimated additive effects of resistance allele for each parent

145

Appendix III: QTL detected via CIM in the RIL populations LMPG-6/Ada (Table A), LMPG-6/Fahari (Table B), LMPG-6/Gem (Table C), LMPG-6/Kudu (Table D), LMPG6/Kulungu (Table E), LMPG-6/Paka (Table F), LMPG-6/Pasa (Table G), LMPG-6/Popo (Table H), and LMPG-6/Romany (Table I). A a

b

c

Add QTL Pos (cM) LOD R2 353.3 4.3 22.5 -4.0 221.6 2.9 16.2 -2.5

Environment Ken13

Chrom Left SNP 4A SNP516 6A SNP425

Right SNP SNP780 SNP5

StP12

2D 6D

SNP219 SNP456

SNP116 SNP194

166.8 75.2

4.0 3.8

8.1 4.8

3.3 2.6

StP13

1B 3B 4A 4B 5D 6A 6B

SNP385 SNP4 SNP891 SNP782 SNP977 SNP425 SNP391

SNP382 SNP813 SNP428 SNP159 SNP491 SNP5 SNP394

10.0 713.8 240.7 393.9 105.6 216.6 407.7

2.6 3.7 3.3 3.4 3.3 2.7 2.9

0.4 3.4 0.7 0.7 7.3 0.0 1.8

-1.0 -1.8 0.8 -1.2 2.1 -0.3 1.6

B a

b

c

Add QTL Pos (cM) LOD R2 215.8 6.1 24.1 2.0

Environment SA12

Chrom Left SNP 2A SNP416

Right SNP SNP767

StP12

2B 4A 4B 5B 7B

SNP294 SNP375 SNP858 SNP471 SNP22

SNP970 SNP90 SNP117 SNP440 SNP421

439.9 196.8 177.3 10.0 350.2

3.0 2.5 2.6 3.8 3.6

5.7 0.0 0.8 2.8 3.1

2.0 0.2 0.7 -1.6 -1.4

StP13

1A 2B 2D 3B 4A 5B

SNP437 SNP294 SNP707 SNP320 SNP375 SNP141

SNP323 SNP970 SNP438 SNP359 SNP90 SNP765

205.5 434.9 275.9 806.0 196.8 575.9

2.8 3.0 2.8 2.5 2.8 2.6

1.0 6.5 1.8 1.7 0.2 1.2

-1.2 3.2 1.8 1.6 0.4 -1.5

146

a

b

c

a

b

c

Left SNP SNP654 SNP60 SNP134

Right SNP SNP697 SNP456 SNP175

Add QTL Pos (cM) LOD R2 338.2 2.9 0.4 0.6 65.7 2.6 2.4 -2.4 281.5 3.2 1.3 -1.0

Environment Ken12

Chrom Left SNP 3B SNP983 7B SNP498

Right SNP SNP982 SNP317

Add QTL Pos (cM) LOD R2 418.8 2.8 12.0 1.9 116.4 2.5 11.0 8.2

StP12

3B

SNP535

SNP866

607.5

3.1

13.1

3.0

StP13

3B 4A

SNP884 SNP383

SNP252 SNP380

441.2 690.1

3.5 3.1

14.5 3.7

2.9 2.9

Environment Ken13

Chrom 2A 3B 4A 4B

Left SNP SNP148 SNP885 SNP371 SNP816

Right SNP SNP744 SNP790 SNP375 SNP895

SA12

4A 6B 7D

SNP445 SNP27 SNP957

SNP402 SNP59 SNP269

268.2 226.6 140.5

3.1 2.9 2.5

9.8 9.8 2.8

1.5 1.0 0.6

StP12

3B 4A 7B

SNP770 SNP380 SNP355

SNP401 SNP200

42.9 714.6 164.9

2.9 2.6 2.8

7.2 5.0 5.5

3.6 2.0 2.9

Environment StP13

Chrom 6A 6D 7B

C

D a

b

c

Add QTL Pos (cM) LOD R2 108.7 3.7 2.8 2.0 684.5 3.5 2.7 1.5 164.3 3.3 1.2 1.5 94.3 2.8 0.1 0.3

E Environment StP13

a

b

c

a

b

c

Chrom Left SNP 6A SNP919

Right SNP SNP506

Add QTL Pos (cM) LOD R2 158.0 2.8 18.4 -1.9

Chrom Left SNP 2B SNP797

Right SNP SNP646

Add QTL Pos (cM) LOD R2 100.5 3.0 10.2 3.8

F Environment StP12

147

G a

b

c

Add QTL Pos (cM) LOD R2 205.7 4.1 8.9 1.2 358.8 4.7 17.4 2.2 850.4 3.1 1.5 0.7

Environment SA12

Chrom 2A 4B 5B

Left SNP SNP580 SNP177 SNP88

Right SNP SNP776 SNP203 SNP192

StP12

6B

SNP220

SNP217

801.3

3.6

13.7

3.7

StP13

2B 6B

SNP294 SNP220

SNP970 SNP217

429.9 806.3

2.7 5.5

12.0 22.0

2.3 3.6

H a

b

c

Add QTL Pos (cM) LOD R2 728.9 3.1 13.5 -3.2 627.4 2.8 10.0 -1.7

Environment Ken13

Chrom Left SNP 5B SNP877 6B SNP500

Right SNP SNP42 SNP441

SA12

5D

SNP714

SNP481

143.0

3.7

15.6

0.7

StP12

5B

SNP142

SNP784

629.2

3.1

12.9

-3.2

I a

b

c

Add QTL Pos (cM) LOD R2 210.3 2.6 10.2 -1.8 136.7 2.9 7.7 -1.5

Environment Ken13

Chrom Left SNP 6A SNP935 6D SNP493

Right SNP SNP425 SNP147

SA12

7B

SNP268

SNP337

43.0

7.6

24.6

4.4

StP12

3B 7A

SNP988 SNP599

SNP824 SNP202

659.4 141.1

2.9 3.0

7.6 8.0

1.8 -2.5

a

LOD values are the peak logarithm of odds score for the given QTL

b

Value indicates the phenotypic variation explained by the QTL

c

Value indicates the estimated additive effect of the QTL; negative value means that the allele was contributed by RB07

148

x 10000

Appendix IV: Distribution of sequence reads by each population and parent (scaled down to 1/10) generated using the GBS approach.

350

300

Number of reads

250

200

Parents

150

100

50

0 Ada

Fahari

Gem

Kudu Kulungu Ngiri

Paka

Pasa

Parents and Populations

149

Popo Romany LMPG-6

Appendix V: A scatter plot of principal component 1 (PC1) plotted against principal component 2 (PC2) of the NAM population.

150

Appendix VI: Frequency distribution of stem rust severity (%) for the ten NAM populations in four environments – St. Paul 2012 (StP12), South Africa 2012 (SA12), St. Paul 2013 (StP13), Kenya 2013 (Ken13).

151

Appendix VI continued

152

Appendix VI continued

153

Appendix VI continued

154

Appendix VI continued

155

Appendix VI continued

156

Appendix VI continued

157

Appendix VI continued

158

0.0 10.0 20.0 25.8 35.8 45.8 55.8 60.1 70.1 80.1 90.1 97.2 107.2 117.2 127.2 137.2 146.1 156.1 160.8 170.8 180.1 190.1 197.2 207.2 214.4 224.4 230.4 240.4 250.4 257.0 267.0 275.9 282.7 292.7 299.6 309.6 319.6 329.0

LOD Value

0.0 17.7 34.9 54.9 70.8 90.5 110.5 128.5 148.5 166.7 184.6 203.8 216.3 232.3 249.0 265.7 280.9 297.0 313.1 330.2 347.3 358.9 376.4 394.7 410.6 429.9 449.8 465.1 484.0 499.9 518.6 534.6 548.4 566.4 583.9 600.0 613.9 632.8

LOD Value 0.0 10.0 20.0 30.0 40.0 50.0 60.0 68.5 78.5 86.1 91.7 101.7 108.7 116.3 121.8 127.2 136.4 141.9 146.3 154.1 159.1 164.9 173.2 178.4 183.5 189.0 194.7 200.7 207.9 215.8 223.1 232.3 240.9 250.9 256.0 266.0 275.8 285.8 295.8 305.3

LOD Value

Appendix VII: Quantitative trait loci (QTL) detected in 11 chromosomes using the iQTLm method of joint mapping approach in four environments. LOD values of the QTL peaks are shown on the Y-axis; positions (cM) of the QTL are shown on X-axis. StP12, SA12, StP13, and Ken13 represent the environments St. Paul 2012, South Africa 2012, St. Paul 2013, and Kenya 2013, respectively. Chromosomes with no QTL detected are not shown. Traits with no significant QTL are not shown. 2A

4

3.5

2.5 3

2 Ken13

1.5 StP12

1 StP13

0.5

0

2B

7

6

5

4 Ken13

3 StP12

2 StP13

1

0

2D

3.5

3

2.5

1.5 2

1 StP12

0.5

0

159

0.0 10.0 20.0 25.8 35.8 45.8 55.8 60.1 70.1 80.1 90.1 97.2 107.2 117.2 127.2 137.2 146.1 156.1 160.8 170.8 180.1 190.1 197.2 207.2 214.4 224.4 230.4 240.4 250.4 257.0 267.0 275.9 282.7 292.7 299.6 309.6 319.6 329.0

LOD Value

0.0 17.7 34.9 54.9 70.8 90.5 110.5 128.5 148.5 166.7 184.6 203.8 216.3 232.3 249.0 265.7 280.9 297.0 313.1 330.2 347.3 358.9 376.4 394.7 410.6 429.9 449.8 465.1 484.0 499.9 518.6 534.6 548.4 566.4 583.9 600.0 613.9 632.8

LOD Value 0.0 10.0 20.0 30.0 40.0 50.0 60.0 68.5 78.5 86.1 91.7 101.7 108.7 116.3 121.8 127.2 136.4 141.9 146.3 154.1 159.1 164.9 173.2 178.4 183.5 189.0 194.7 200.7 207.9 215.8 223.1 232.3 240.9 250.9 256.0 266.0 275.8 285.8 295.8 305.3

LOD Value

Appendix VII continued 2A

4

3.5

2.5 3

Ken13

1.5 2

StP12

1 StP13

0.5

0

2B

7

6

5

4 Ken13

3 StP12

2 StP13

1

0

2D

3.5

3

2.5

2

1.5

1 StP12

0.5

0

160

0.0 37.3 70.9 113.0 156.9 184.1 215.3 239.7 260.3 280.5 296.1 312.2 334.9 348.3 363.9 375.8 388.9 399.5 407.3 414.7 420.4 429.3 434.4 437.4 443.3 452.1 459.3 470.5 483.1 497.9 509.6 526.8 542.9 559.7 579.2 607.0 638.7 673.1 711.0

LOD Value 0.0

161 267.4

260.8

250.8

240.8

230.8

222.2

212.2

206.6

196.6

186.6

180.1

170.1

160.1

150.1

143.0

133.2

126.0

116.0

110.6

100.4

90.4

83.4

73.4

66.4

56.4

46.4

40.0

30.0

20.0

10.0

LOD Value 0.0 29.3 59.3 89.0 114.4 142.3 170.7 197.5 223.9 249.3 275.7 303.8 333.4 362.2 390.3 419.7 445.5 467.3 493.4 518.0 540.9 568.3 592.0 615.8 642.5 668.3 696.6 723.9 750.9 777.2 804.7 829.4 855.4 883.0 905.4 935.4 956.6 986.6

LOD Value

Appendix VII continued 5B

4.5

3.5 4

2.5 3 Ken13

2 StP12

1.5

1 StP13

0.5

0

5D

3.5

3

2.5

2

1.5 StP12

1 StP13

0.5

0

6A

4.5

3.5 4

2.5 3

1.5 2

Ken13

0.5 1

0

0.0 15.0 28.8 43.0 55.7 70.7 79.8 93.6 108.6 120.9 133.8 147.9 154.9 165.1 179.6 190.6 203.2 216.0 226.7 239.2 252.1 263.4 275.2 286.5 298.7 313.6 328.3 343.3 355.2 369.4 384.4 394.7 407.8

LOD Value LOD Value

Appendix VII continued 7A

4.5

3.5 4

2.5 3

1.5 2

Ken13

0.5 1

0

7B

9

8

7

6

5

4

3 SA12

2

1

0

162

Appendix VIII: Field reaction to stem rust observed on the RB07/MN06113-8 RIL population. Environment Kenya 2012 Line Severitya IRb MN06_1 17.5 MS MN06_2 35.0 MSS MN06_4 10.0 MS MN06_5 17.5 MSMR MN06_6 15.0 MS MN06_7 20.0 MS MN06_8 25.0 MSMR MN06_9 5.0 MRMS MN06_10 30.0 MS MN06_11 10.0 MRMS MN06_12 45.0 MS MN06_13 5.0 MRMS MN06_14 25.0 MS MN06_15 17.5 MRMS MN06_16 40.0 MSMR MN06_17 2.5 MS MN06_18 47.5 MSS MN06_20 50.0 MS MN06_21 25.0 MRMS MN06_22 25.0 MRMS MN06_23 17.5 MSS MN06_24 15.0 MRMS MN06_25 30.0 MRMS MN06_26 10.0 MRMS MN06_27 27.5 MRMS MN06_28 20.0 MRMS MN06_29 5.0 MRMS MN06_30 7.5 MRMS MN06_31 30.0 MRMS MN06_33 37.5 MRMS MN06_34 20.0 MRMS MN06_35 25.0 MRMS MN06_36 17.5 MRMS MN06_37 12.5 MRMS MN06_38 17.5 MS Environment Kenya 2012

Kenya 2013 Ethiopia 2013 St. Paul 2013 a b a b Severity IR Severity IR Severitya IRb 25.0 MRMS 10.0 MRMS 35.0 MS 20.0 MRMS 50.0 S 20.0 RMR 35.0 MRMS 15.0 MS 10.0 MRMS 35.0 MS 60.0 S 40.0 MSS 25.0 MS 40.0 MSMR 35.0 MSS 60.0 S 35.0 MSS 45.0 MSS 20.0 RMR 50.0 S 40.0 MS 20.0 R 60.0 S 25.0 MS 30.0 MS 15.0 MRMS 10.0 MRMS 70.0 S 40.0 MS 60.0 S 20.0 RMR 35.0 30.0 MRMS 45.0 MSS 50.0 S 25.0 MRMS 40.0 MSMR 20.0 MRMS 37.5 MSS 60.0 S 20.0 MRMS 15.0 MS 15.0 MRMS 60.0 S 15.0 RMR 27.5 MRMS 30.0 MRMS 10.0 RMR 15.0 R 37.5 MSS 40.0 MSS 10.0 RMR 27.5 MSMR 40.0 MRMS 15.0 MRMS 60.0 S 35.0 MS 22.5 MSS 45.0 MSS 15.0 MRMS 60.0 S 20.0 MRMS 60.0 S 30.0 MSS 15.0 S 25.0 MRMS 45.0 MSS 20.0 MRMS Kenya 2013 Ethiopia 2013 St. Paul 2013

163

Line Severitya IRb MN06_39 1.0 MRMS MN06_40 12.5 MRMS MN06_41 42.5 MRMS MN06_42 47.5 MS MN06_43 42.5 MRMS MN06_44 15.0 MRMS MN06_45 20.0 MSMR MN06_46 40.0 MRMS MN06_47 15.0 MS MN06_49 50.0 S MN06_50 27.5 MS MN06_51 30.0 MRMS MN06_52 15.0 MS MN06_53 7.5 MRMS MN06_54 5.0 MRMS MN06_55 5.0 MS MN06_56 45.0 MRMS MN06_57 7.5 MRMS MN06_58 27.5 MS MN06_59 22.5 MRMS MN06_60 12.5 MRMS MN06_61 17.5 MRMS MN06_62 37.5 MRMS MN06_63 15.0 MRMS MN06_64 12.5 MRMS MN06_65 37.5 MRMS MN06_66 27.5 MS MN06_67 50.0 MSMR MN06_68 1.0 MRMS MN06_69 42.5 MSS MN06_70 35.0 MSMR MN06_71 7.5 MRMS MN06_72 5.0 MRMS MN06_73 40.0 MRMS MN06_74 7.5 MSS MN06_75 45.0 MS MN06_76 40.0 MSMR MN06_78 15.0 MRMS MN06_79 20.0 S Environment Kenya 2012

Severitya IRb Severitya IRb Severitya IRb 15.0 MRMS 35.0 S 10.0 R 50.0 SMS 25.0 MRMS 60.0 S 30.0 MS 60.0 S 20.0 R 15.0 MRMS 15.0 MRMS 60.0 SMS 20.0 R 40.0 S 40.0 S 10.0 MRMS 37.5 S 55.0 SMS 25.0 R 35.0 MS 37.5 S 60.0 S 15.0 R 17.5 MSS 40.0 MSS 15.0 MRMS 25.0 MS 15.0 R 35.0 MSS 50.0 MSS 10.0 MRMS 50.0 SMS 20.0 MRMS 45.0 S 20.0 MRMS 27.5 MSMR 35.0 MSMR 15.0 R 30.0 MS 20.0 S 15.0 R 40.0 S 10.0 MRMS 15.0 MRMS 30.0 MS 10.0 MRMS 30.0 MS 45.0 MSS 20.0 MRMS 32.5 MSS 40.0 SMS 20.0 MRMS 37.5 SMS 30.0 MSS 20.0 R 37.5 MSS 35.0 MS 20.0 SMS 45.0 MSS 20.0 MRMS 40.0 SMS 55.0 SMS 25.0 MS 17.5 S 40.0 S 20.0 MRMS 20.0 MRMS 50.0 MS 15.0 MRMS 40.0 MSS 15.0 MRMS 12.5 MRMS 35.0 MS 20.0 MRMS 45.0 S 60.0 S 25.0 MSS 22.5 S 25.0 S 15.0 MRMS 42.5 SMS 70.0 S 20.0 MRMS 40.0 SMS 10.0 R 10.0 R 15.0 MRMS Kenya 2013 Ethiopia 2013 St. Paul 2013

164

Line Severitya IRb MN06_80 7.5 MRMS MN06_81 42.5 MSMR MN06_82 37.5 MS MN06_83 37.5 MS MN06_84 32.5 MSMR MN06_85 37.5 MS MN06_86 37.5 MSMR MN06_87 35.0 MS MN06_88 7.5 MRMS MN06_89 10.0 MRMS MN06_91 35.0 MS MN06_93 32.5 MRMS MN06_94 35.0 MS MN06_95 7.5 MRMS MN06_96 15.0 MRMS MN06_97 30.0 MRMS MN06_98 17.5 MS MN06_99 37.5 MRMS MN06_100 17.5 MRMS MN06_101 10.0 MRMS MN06_102 35.0 MS MN06_103 37.5 MSMR MN06_104 7.5 MRMS MN06_105 7.5 MRMS MN06_106 12.5 MSS MN06_107 22.5 S MN06_108 17.5 MRMS MN06_109 17.5 MSMR MN06_110 47.5 MRMS MN06_111 27.5 MSMR MN06_112 22.5 MSMR MN06_113 17.5 MS MN06_114 47.5 MSMR MN06_115 42.5 MSMR MN06_116 35.0 MRMS MN06_117 40.0 MRMS MN06_118 50.0 MS MN06_119 10.0 MRMS MN06_120 37.5 MSMR Environment Kenya 2012

Severitya IRb Severitya IRb Severitya IRb 27.5 MSMR 15.0 MRMS 60.0 MS 15.0 R 25.0 S 50.0 SMS 20.0 MRMS 45.0 SMS 25.0 MS 45.0 S 60.0 S 30.0 MS 25.0 MRMS 40.0 SMS 30.0 MS 30.0 S 55.0 S 25.0 MRMS 15.0 R 15.0 R 17.5 MSS 20.0 MRMS 60.0 S 30.0 MS 50.0 S 20.0 MRMS 30.0 MS 45.0 S 25.0 R 45.0 MSS 20.0 MRMS 32.5 MS 45.0 S 30.0 MS 37.5 SMS 60.0 S 35.0 MSS 25.0 MR 15.0 R 20.0 MRMS 20.0 S 50.0 S 25.0 MRMS 30.0 MSMR 35.0 MS 25.0 MRMS 50.0 S 20.0 R 20.0 R 27.5 MSS 40.0 S 25.0 MS 5.0 MS 20.0 MRMS 27.5 MSS 45.0 MSS 15.0 MRMS 40.0 MSS 45.0 S 25.0 MRMS 30.0 MS 20.0 MRMS 20.0 MRMS 40.0 S 20.0 MRMS 20.0 R 25.0 MRMS 15.0 R 15.0 R 60.0 S 20.0 MRMS Kenya 2013 Ethiopia 2013 St. Paul 2013

165

Line MN06_121 MN06_122 MN06_123 MN06_124 MN06_125 MN06_126 MN06_127 MN06_128 MN06_129 MN06_130 MN06_131 MN06_132 MN06_133 MN06_134 MN06_135 MN06_137 MN06_138 MN06_139 MN06_140 MN06_141 MN06_142 MN06_143 MN06_144 MN06_145 MN06_146 MN06_147 MN06_148 MN06_149 a

Severitya 22.5 20.0 1.0 7.5 5.0 37.5 2.5 30.0 2.5 5.0 37.5 22.5 35.0 35.0 17.5 20.0 10.0 17.5 12.5 42.5 30.0 45.0 35.0 32.5 22.5 7.5 17.5 20.0

IRb MRMS MRMS S MS MSMR MSMR S MSMR MS MSMR MRMS MRMS MSMR MSS MRMS MRMS MRMS MRMS MRMS MSS MS MS MS MS MR MRMS MRMS MRMS

Severitya 20.0 17.5 27.5 -

IRb S MS MS -

Severitya IRb Severitya 15.0 60.0 MSS 35.0 10.0 20.0 30.0 MRMS 15.0 20.0 15.0 30.0 15.0 15.0 20.0 60.0 S 20.0 55.0 SMS 30.0 15.0 20.0 25.0 15.0 15.0 40.0 MS 10.0 20.0 25.0 15.0 20.0 50.0 MSS 20.0 20.0 20.0 55.0 S 15.0 60.0 S 20.0

Disease severity (%) was measured according to modified Cobb scale (Peterson et al.

1948) b

IRb MRMS MSS MS MRMS MRMS MRMS MRMS MS MS MS MRMS MRMS MS MS MRMS MRMS MS MS MS MRMS MRMS MRMS MRMS MRMS MRMS MRMS MRMS R

Represents infection response of the plants to stem rust

166

Appendix IX: Field reaction to stem rust observed on the 10 NAM RIL populations.

Environment Line Ada_1 Ada_2 Ada_3 Ada_5 Ada_6 Ada_7 Ada_8 Ada_9 Ada_11 Ada_12 Ada_13 Ada_14 Ada_15 Ada_17 Ada_18 Ada_19 Ada_21 Ada_22 Ada_23 Ada_24 Ada_26 Ada_27 Ada_28 Ada_29 Ada_30 Ada_31 Ada_32 Ada_33 Ada_34 Ada_35 Ada_37 Ada_38 Ada_41 Ada_42 Ada_43 Ada_44

St. Paul 2012 Severitya IRb 55 S 40 S 20 MSS 25 MSS 65 S 30 MSS 55 MR 45 MRMSS 15 MR 15 MSS 30 MSS 60 S 60 S 40 MSS 35 MSS 45 MSS 55 S 40 MSS 45 MSS 55 S 55 S 25 MSS 45 S 50 S 15 RMR 50 S 60 S 60 S 35 S 60 S 45 MSS 35 S 50 S 65 S 60 S 20 MSS

South Africa 2012 Severitya IRb 3 4 4 2 3 1 2 6 3 6 2 4 2 2 2 4 2 8 3 5 9 3 6 9 3 9 4 5 8 3 2 2 3 2 3 1 -

167

St. Paul 2013 Severitya IRb 90 S 45 MSS 20 MSS 50 S 60 S 25 MSS 65 S 55 S 30 MRMS 25 MRMS 35 MSS 50 MSS 45 MRMS 15 MRMS 40 MSS 45 MSS 75 S 65 S 55 MSS 55 MSS 50 S 15 MRMS 45 MRMS 55 MSS 15 MRR 50 MRMS 40 MSS 50 MSS 35 MSS 55 S 50 MSS 10 MSS 50 MSS 45 MSS 35 MRMS 10 MSS

Kenya 2013 Severitya IRb 20 MSS 5 MSS 5 MSS 30 S 50 MSS 15 MS 45 S 15 M 15 M 1 R 30 S 40 MSS 30 M 10 MSS 1 R 30 MSS 30 S 10 MS 15 M 15 M 20 MSS 15 MSS 20 S 55 S 40 MSS 55 S 10 M 20 M 20 MSS 40 S 30 S 20 MSS 40 MSS 20 M 30 M 20 M

Environment Line Ada_45 Ada_46 Ada_47 Ada_48 Ada_49 Ada_50 Ada_51 Ada_52 Ada_54 Ada_56 Ada_57 Ada_58 Ada_59 Ada_61 Ada_62 Ada_63 Ada_64 Ada_65 Ada_66 Ada_67 Ada_68 Ada_69 Ada_70 Ada_71 Ada_72 Ada_73 Ada_74 Ada_75 Ada_76 Ada_77 Ada_78 Ada_79 Ada_81 Ada_82 Ada_84 Fahari_1 Fahari_2 Fahari_3 Fahari_4

St. Paul 2012 Severitya IRb 25 MS 20 MSS 55 S 65 S 45 S 60 S 45 S 45 MSS 20 M 60 S 45 MSS 25 S 20 M 65 S 25 MSS 40 MSS 30 MSS 65 S 50 MSS 60 S 45 MSS 50 S 50 S 45 S 35 MSS 60 S 40 MSS 70 S 40 MSS 40 S 45 S 45 M 20 RMR 40 MSS 45 S 30 MSS 20 RMR 30 MSS 35 MSS

South Africa 2012 Severitya IRb 1 5 3 2 5 6 6 6 6 7 9 3 6 7 4 3 4 2 2 2 2 2 9 8 7 8 6 8 8 7 6 5 8 7 8 8 3 6 5 -

168

St. Paul 2013 Severitya IRb 25 MRMS 10 MR 35 MRMS 55 MSS 55 MSS 50 MSS 40 S 40 S 20 MRMS 30 MRMS 25 MRMS 20 MSS 25 MRMS 30 MRMS 40 S 25 MR 30 MSS 50 MSS 60 MSS 25 S 45 MSS 45 MSS 35 MRMS 20 S 40 MSS 50 S 20 MRMS 55 MSS 25 MSS 25 MSS 30 MRMS 30 MRMS 25 MR 40 S 25 RMR 10 MSS 15 RMR 25 MSS 25 MRMS

Kenya 2013 Severitya IRb 30 MSS 20 S 40 S 40 MSS 15 M 30 MSS 30 MSS 15 M 10 M 15 M 30 MSS 10 MSS 25 MSS 20 MSS 30 MR 40 MSS 20 S 40 MSS 45 S 30 S 40 M 15 M 25 S 15 MSS 15 M 45 S 10 MSS 40 MSS 20 S 20 MSS 35 M 40 MSS 15 M 30 MSS 20 MSS 1 R 1 R 5 RMR 10 M

Environment Line Fahari_5 Fahari_6 Fahari_7 Fahari_8 Fahari_9 Fahari_10 Fahari_12 Fahari_13 Fahari_14 Fahari_16 Fahari_18 Fahari_19 Fahari_22 Fahari_23 Fahari_25 Fahari_26 Fahari_27 Fahari_28 Fahari_29 Fahari_30 Fahari_31 Fahari_32 Fahari_33 Fahari_34 Fahari_35 Fahari_36 Fahari_37 Fahari_38 Fahari_40 Fahari_41 Fahari_42 Fahari_43 Fahari_44 Fahari_45 Fahari_46 Fahari_47 Fahari_48 Fahari_49 Fahari_50

St. Paul 2012 Severitya IRb 35 M 35 M 20 MR 10 MSS 20 MSS 30 M 30 RMRMS 40 MSS 40 M 55 S 20 RMR 20 RMR 15 RMR 30 RMRMS 30 MS 20 RMR 20 MSS 30 M 25 RMR 40 MS 20 MR 30 M 20 MSS 30 MS 25 MS 40 MSS 30 M 20 M 20 RMR 25 M 20 RMR 30 MSS 35 MS 15 RMR 40 MSS 35 MSS 45 S 20 MSS 30 M

South Africa 2012 Severitya IRb 4 3 2 2 6 7 4 4 6 4 5 5 7 6 2 4 4 4 4 5 3 5 3 4 4 5 3 6 4 6 4 5 6 5 8 3 4 4 6 -

169

St. Paul 2013 Severitya IRb 20 MRMS 40 MRMS 25 MRMS 10 RMR 10 RMR 10 RMR 45 MRMS 30 MRMS 20 MRMS 80 S 10 R 25 RMR 10 RMR 25 MRMS 35 MRMS 20 RMR 5 RMR 30 MRMS 25 RMR 65 S 25 RMR 30 MRMS 30 MRMS 30 MRMS 45 MRMS 35 MSS 40 MSS 30 MRMS 20 RMR 25 RMR 25 MRR 5 MRMS 30 MRMS 15 R 30 MSS 20 MRMS 55 S 10 S 40 MRMS

Kenya 2013 Severitya IRb 10 MSS 10 M 1 R 1 R 5 MSS 5 MS 10 M 1 S 5 MS 5 M 1 S 1 R 1 S 5 M 5 M 1 R 5 MS 5 M 25 M 5 RMR 20 M 5 R 1 R 20 MSS 5 RMR 10 MSS 1 MS 10 MS 20 MR 5 M 10 MSS 1 S 15 MSS 5 M 5 MSS 5 MSS 20 M 5 MS 10 M

Environment Line Fahari_51 Fahari_52 Fahari_53 Fahari_54 Fahari_55 Fahari_57 Fahari_58 Fahari_59 Fahari_60 Fahari_61 Fahari_62 Fahari_63 Fahari_64 Fahari_65 Fahari_66 Fahari_67 Fahari_68 Fahari_69 Fahari_70 Fahari_71 Fahari_72 Fahari_73 Fahari_76 Fahari_77 Fahari_78 Fahari_79 Fahari_80 Fahari_81 Fahari_83 Fahari_85 Fahari_86 Fahari_87 Fahari_88 Fahari_89 Fahari_90 Fahari_91 Fahari_92 Fahari_93 Fahari_94

St. Paul 2012 Severitya IRb 15 M 35 MSS 45 S 20 RMR 20 RMR 25 M 30 MS 35 MSS 35 MS 40 M 30 MSS 35 MSS 30 RMRMS 20 RMR 35 MSS 20 M 15 MSS 50 MSS 45 MSS 50 S 35 M 25 RMR 30 MSS 20 RMR 35 MS 45 MSS 40 MSS 35 MSS 35 MSS 30 M 35 MS 25 RMR 40 MS 50 S 20 RMR 35 MSS 40 MS 45 S 20 RMR

South Africa 2012 Severitya IRb 6 3 4 3 5 8 5 4 5 9 4 4 5 4 3 7 1 4 9 4 6 5 5 9 4 4 5 5 5 3 4 2 4 4 4 9 5 -

170

St. Paul 2013 Severitya IRb 5 S 35 MSS 60 MSS 45 MRMS 20 RMR 15 RMR 30 MSS 25 MRMS 40 MRMS 35 MSS 25 S 30 MSS 20 MRMS 15 MRMS 25 MSS 30 RMR 5 MR 30 MSS 60 MSS 45 MRMS 45 MRMS 25 MRMS 20 MRMS 15 RMR 40 MRMS 40 MRMS 30 MRMS 10 S 55 S 40 MRMS 30 MRMS 25 RMR 20 MSS 50 S 45 S 50 S 55 S 60 MSS 20 RMR

Kenya 2013 Severitya IRb 5 MS 10 MSS 15 M 10 MSS 5 M 5 M 1 R 1 R 10 M 15 MSS 10 MSS 10 M 10 M 10 M 5 M 5 M 1 S 5 M 20 MSS 25 M 5 M 15 M 0 R 5 RMR 1 R 15 MS 10 M 5 MSS 10 MSS 10 M 10 RMR 20 M 5 MSS 15 M 5 M 5 M 20 M 10 M 10 MSS

Environment Line Fahari_95 Fahari_96 Fahari_97 Fahari_98 Fahari_99 Fahari_100 Fahari_101 Fahari_102 Gem_1 Gem_2 Gem_3 Gem_4 Gem_6 Gem_7 Gem_8 Gem_9 Gem_10 Gem_11 Gem_12 Gem_13 Gem_14 Gem_15 Gem_16 Gem_17 Gem_18 Gem_19 Gem_20 Gem_21 Gem_22 Gem_23 Gem_24 Gem_25 Gem_26 Gem_27 Gem_28 Gem_29 Gem_30 Gem_31 Gem_32

St. Paul 2012 Severitya IRb 15 RMR 30 M 30 M 25 MSS 45 MSS 35 MSS 35 M 25 M 40 MSS 45 S 60 S 40 MSS 35 MSS 40 S 50 MSS 20 RMR 35 MSS 65 S 55 S 20 MSS 45 MSS 40 M 40 MSS 25 M 45 S 40 MS 40 MSS 40 MSS 35 M 15 RMR 20 RMR 45 MS 25 M 40 MSS 40 MSS 40 MRMSS 50 S 25 M 50 S

South Africa 2012 Severitya IRb 8 5 5 4 5 3 6 7 9 5 6 10 8 7 8 7 6 3 5 4 8 5 8 8 6 5 5 5 9 5 5 4 3 7 5 3 6 -

171

St. Paul 2013 Severitya IRb 15 RMR 30 MRMS 40 RMR 10 MSS 50 MSS 20 MSS 35 MRMS 30 MRMS 40 MRMS 15 MRMS 75 S 15 MRMS 30 MSS 45 S 50 S 25 MSS 45 MSS 50 S 35 MSS 30 MSS 25 MSS 60 S 45 S 25 MRMS 55 S 40 S 40 S 30 MRMS 50 S 50 S 35 MRMS 45 S 40 MRMS 40 MRMS 40 MRMS 35 MRMS 45 MSS 40 MRMS 50 S

Kenya 2013 Severitya IRb 10 MSS 20 MSS 15 MS 5 MSS 40 S 1 S 5 MS 1 S 30 S 5 MS 40 MSS 1 R 10 MSS 10 MSS 20 MSS 1 MS 10 MSS 35 M 30 S 10 MSS 10 MS 15 MSS 20 MSS 1 R 35 MSS 10 M 10 M 1 S 25 MSS 5 MS 5 MSS 30 S 20 M 10 MSS 10 M 5 M 20 MSS 30 MSS 35 S

Environment Line Gem_33 Gem_34 Gem_35 Gem_36 Gem_37 Gem_39 Gem_40 Gem_41 Gem_42 Gem_43 Gem_44 Gem_45 Gem_46 Gem_47 Gem_48 Gem_49 Gem_50 Gem_51 Gem_52 Gem_53 Gem_54 Gem_55 Gem_56 Gem_57 Gem_58 Gem_59 Gem_60 Gem_62 Gem_63 Gem_65 Gem_66 Gem_67 Gem_68 Gem_69 Gem_70 Gem_71 Gem_72 Gem_73 Gem_74

St. Paul 2012 Severitya IRb 65 S 20 MSS 45 S 55 S 50 S 50 S 40 MSS 25 M 20 MS 30 MS 40 MSS 35 S 35 S 45 S 45 S 25 MS 40 S 30 MRMSS 20 RMR 50 S 45 MSS 45 MSS 45 MSS 15 MSS 30 M 20 M 35 S 40 S 25 M 45 S 45 S 20 RMR 45 S 40 MS 50 MSS 55 S 55 MSS 45 S 40 MSS

South Africa 2012 Severitya IRb 6 9 5 5 4 4 6 7 6 4 5 7 5 4 4 3 6 10 5 6 5 6 5 6 5 8 8 6 5 5 5 4 7 4 8 5 6 -

172

St. Paul 2013 Severitya IRb 45 MS 10 MSS 45 MRMS 60 S 60 S 40 MRMS 50 S 10 RMR 20 MSS 25 MRMS 35 MRMS 30 MSS 25 MRMS 40 MSS 45 MSS 35 MSS 30 MSS 50 MRMS 20 MRMS 50 MRMS 40 MRMS 45 MRMS 30 MSS 10 MRMS 35 MSS 10 MSS 25 MSS 40 MRMS 45 MRMS 45 MRMS 25 S 25 RMR 40 MRMS 30 MRMS 40 MRMS 45 MRMS 45 MSS 40 MSS 30 MRMS

Kenya 2013 Severitya IRb 5 MSS 1 R 5 MSS 30 MSS 20 MSS 10 MSS 30 MSS 0 R 5 MSS 1 MS 5 MS 10 MS 1 RMR 10 MSS 25 MSS 1 S 10 MSS 10 M 5 M 10 M 30 MSS 15 M 15 M 5 M 1 R 1 R 10 MSS 10 MSS 5 M 15 MS 5 MS 15 M 30 MSS 20 MSS 1 R 20 M 35 MSS 30 MSS 10 MSS

Environment Line Gem_75 Gem_77 Gem_78 Gem_79 Gem_80 Gem_81 Gem_82 Gem_83 Gem_84 Gem_85 Gem_86 Gem_87 Gem_88 Gem_90 Gem_91 Gem_92 Gem_93 Gem_94 Gem_95 Gem_96 Gem_98 Gem_99 Gem_100 Gem_101 Gem_102 Gem_103 Gem_104 Kudu_1 Kudu_2 Kudu_3 Kudu_4 Kudu_5 Kudu_6 Kudu_7 Kudu_8 Kudu_9 Kudu_10 Kudu_11 Kudu_12

St. Paul 2012 Severitya IRb 55 S 50 S 55 MSS 40 MSS 25 M 45 S 60 S 35 MSS 20 RMR 35 MSS 20 RMR 20 RMR 35 M 35 MS 45 MSS 20 RMR 45 MSS 35 MSS 45 S 60 S 55 S 65 S 50 S 25 RMR 50 S 20 RMR 30 M 15 MR 20 RMR 20 RMR 35 M 40 S 35 MS 35 MS 20 RMR 15 RMR 30 M 35 M 45 S

South Africa 2012 Severitya IRb 3 4 7 4 4 7 6 6 5 5 3 4 8 3 4 5 6 5 7 6 4 6 5 6 4 5 6 6 6 7 6 8 9 7 6 5 9 8 -

173

St. Paul 2013 Severitya IRb 50 S 45 MSS 55 S 35 MSS 25 MSS 45 MSS 55 S 25 S 25 MRMS 25 MRMS 15 MRMS 30 RMR 30 MRMS 30 MSS 50 S 30 RMR 45 S 35 MRMS 45 S 40 MSS 35 MSS 45 MRMS 35 MRMS 50 S 40 MRMS 30 MRMS 25 MRMS 10 MRMS 25 RMR 20 MRMS 35 MRMS 35 MRMS 35 MRMS 35 MRMS 20 RMR 20 RMR 30 MSS 40 MSS 35 MRMS

Kenya 2013 Severitya IRb 35 MSS 10 MSS 30 M 30 MSS 5 MS 60 S 50 S 15 MSS 1 R 20 MSS 20 MSS 5 MR 10 M 1 R 25 MSS 5 MR 20 M 1 R 15 MSS 20 M 5 MS 30 S 10 MSS 25 MSS 10 MSS 20 MSS 10 M 10 M 5 M 5 M 20 M 30 S 10 MSS 40 MSS 5 M 1 R 5 MSS 20 MSS 30 S

Environment Line Kudu_13 Kudu_14 Kudu_15 Kudu_16 Kudu_17 Kudu_19 Kudu_20 Kudu_21 Kudu_22 Kudu_23 Kudu_24 Kudu_25 Kudu_26 Kudu_28 Kudu_29 Kudu_30 Kudu_31 Kudu_32 Kudu_33 Kudu_34 Kudu_35 Kudu_37 Kudu_38 Kudu_39 Kudu_40 Kudu_43 Kudu_44 Kudu_45 Kudu_47 Kudu_48 Kudu_50 Kudu_51 Kudu_52 Kudu_53 Kudu_54 Kudu_55 Kudu_56 Kudu_57 Kudu_58

St. Paul 2012 Severitya IRb 10 MR 25 M 25 M 40 MSS 35 MS 25 M 30 M 20 RMR 15 RMR 40 MS 30 MR 20 RMR 15 RMR 35 MSS 20 RMR 10 RMR 40 S 15 RMR 55 S 30 MS 30 MR 55 S 35 MSS 20 MR 20 MR 10 RMR 25 RMR 45 S 10 RMR 40 S 20 M 15 RMR 30 M 55 S 35 MSS 30 M 30 MS 45 MSS 15 RMR

South Africa 2012 Severitya IRb 7 6 7 6 5 4 5 5 6 5 6 5 6 8 8 8 7 7 5 8 7 7 5 5 5 4 5 6 6 7 5 5 4 5 4 4 -

174

St. Paul 2013 Severitya IRb 10 MSS 15 MSS 30 MRMS 20 MRMS 30 RMR 45 S 50 S 25 MRMS 25 MRMS 60 S 30 MRMS 20 MRMS 30 MRMS 30 MRMS 10 MRMS 10 MRMS 50 S 10 RMR 45 MSS 35 MSS 30 MSS 45 S 30 MSS 30 MSS 15 RMR 20 MRMS 30 MRMS 55 S 25 MRMS 20 MRMS 20 MRMS 15 MRMS 20 MRMS 45 S 25 MRMS 30 MSS 35 MSS 30 MRMS 10 RMR

Kenya 2013 Severitya IRb 5 RMR 10 MSS 1 R 10 M 20 M 45 S 15 M 0 R 1 RMR 25 M 5 RMR 5 MR 0 R 30 MSS 1 R 1 MS 40 S 1 R 25 MSS 25 M 10 MSS 55 S 25 MSS 1 S 10 MSS 1 R 15 M 30 MSS 1 R 30 S 5 M 1 R 40 S 25 MSS 40 S 1 R 15 M 40 S 0 R

Environment Line Kudu_59 Kudu_60 Kudu_61 Kudu_62 Kudu_63 Kudu_64 Kudu_65 Kudu_66 Kudu_67 Kudu_69 Kudu_70 Kudu_71 Kudu_72 Kudu_73 Kudu_74 Kudu_75 Kudu_76 Kudu_77 Kudu_79 Kudu_80 Kudu_81 Kudu_82 Kudu_83 Kudu_84 Kudu_85 Kudu_86 Kudu_87 Kudu_88 Kudu_89 Kulungu_1 Kulungu_3 Kulungu_4 Kulungu_5 Kulungu_6 Kulungu_7 Kulungu_8 Kulungu_9 Kulungu_10 Kulungu_12

St. Paul 2012 Severitya IRb 50 S 0 R 50 S 15 RMR 55 S 55 S 40 MSS 40 MS 15 MSS 15 RMR 20 RMRMS 30 MS 15 RMR 35 M 35 MS 45 MSS 55 S 40 S 40 S 35 MSS 30 MS 10 RMR 30 MSS 20 M 15 M 15 RMR 20 RMR 10 RMR 20 M 30 MS 15 MS 20 MS 65 S 15 RMR 15 RMR 20 S 30 M 40 MSS 20 M

South Africa 2012 Severitya IRb 8 6 8 5 4 4 5 5 6 4 4 5 4 4 5 5 5 5 6 6 6 6 6 6 5 6 7 3 5 4 7 7 7 5 5 6 7 6 7 -

175

St. Paul 2013 Severitya IRb 30 MRMS 15 MRMS 25 MSS 5 RMR 55 S 35 MSS 35 MSS 30 MSS 25 MSS 25 MRMS 25 MRMS 20 MRMS 15 RMR 30 MRMS 20 RMR 30 MRMS 35 MSS 15 MRMS 20 MRMS 15 MRMS 20 MRMS 15 MRMS 20 MRMS 15 MRMS 20 MRMS 15 RMR 15 RMR 15 RMR 20 RMR 15 RMR 10 R 10 R 20 MSS 15 MRMS 15 MRMS 20 MRMS 15 RMR 20 MSS 10 RMR

Kenya 2013 Severitya IRb 40 S 5 MSS 10 M 1 R 30 MSS 25 MSS 20 MSS 30 MSS 1 R 1 R 5 M 20 MSS 5 M 5 M 10 MSS 30 MSS 30 MSS 20 MSS 30 MSS 10 MSSS 5 M 0 R 10 MSS 25 MSS 10 M 5 M 1 R 1 R 5 M 30 MSS 5 MS 15 MSS 60 S 5 MSS 10 MSS 1 S 10 M 40 S 1 M

Environment Line Kulungu_13 Kulungu_15 Kulungu_17 Kulungu_18 Kulungu_19 Kulungu_20 Kulungu_21 Kulungu_22 Kulungu_23 Kulungu_24 Kulungu_25 Kulungu_26 Kulungu_27 Kulungu_28 Kulungu_29 Kulungu_30 Kulungu_31 Kulungu_32 Kulungu_35 Kulungu_36 Kulungu_37 Kulungu_39 Kulungu_42 Kulungu_43 Kulungu_44 Kulungu_45 Kulungu_46 Kulungu_47 Kulungu_48 Kulungu_49 Kulungu_50 Kulungu_51 Kulungu_52 Kulungu_53 Kulungu_54 Kulungu_55 Kulungu_56 Kulungu_57 Kulungu_58

St. Paul 2012 Severitya IRb 15 M 35 MS 30 MS 15 MSS 25 MSS 20 MSS 30 M 10 MS 30 MS 15 MS 30 MSS 20 M 35 S 20 M 30 MSS 35 MSS 25 M 20 M 35 MSS 15 M 15 MS 15 M 35 M 20 MS 40 S 20 S 20 M 30 S 10 MS 20 RMR 25 MS 30 M 40 S 40 S 20 M 15 M 30 MS 50 S 40 S

South Africa 2012 Severitya IRb 7 6 7 6 5 7 7 6 5 8 5 5 8 4 8 5 6 4 5 7 7 7 5 6 4 6 8 5 9 8 6 5 7 7 9 7 5 5 -

176

St. Paul 2013 Severitya IRb 10 RMR 20 MSS 25 MSS 15 MSS 20 MRMS 20 MSS 20 MRMS 5 R 20 MRMS 5 MS 15 MSS 10 MR 25 MSS 15 MRMS 15 MRMS 15 MSS 20 MRMS 20 RMR 15 MRMS 10 RMR 10 RMR 15 MSS 25 S 10 RMR 25 MSS 15 MRMS 10 RMR 15 MRMS 15 MRMS 15 RMR 25 MSS 10 RMR 25 MRMS 20 MSS 10 RMR 10 RMR 20 MRMS 25 MSS 20 MRMS

Kenya 2013 Severitya IRb 5 MS 20 MSS 10 MSS 5 MS 5 M 1 MR 20 MSS 10 M 25 MSS 5 M 30 MSS 5 M 10 MSS 1 S 5 M 5 M 5 M 10 MSS 10 MSS 5 MSS 0 R 5 M 30 S 20 MSS 5 M 20 MSS 5 M 40 S 1 R 15 M 5 MS 5 MSS 5 M 20 M 10 M 0 R 5 M 30 S 10 MSS

Environment Line Kulungu_59 Kulungu_62 Kulungu_65 Kulungu_66 Kulungu_67 Kulungu_68 Kulungu_69 Kulungu_70 Kulungu_71 Kulungu_72 Ngiri_2 Ngiri_3 Ngiri_4 Ngiri_7 Ngiri_8 Ngiri_9 Ngiri_10 Ngiri_11 Ngiri_12 Ngiri_13 Ngiri_14 Ngiri_15 Ngiri_16 Ngiri_17 Ngiri_18 Ngiri_19 Ngiri_21 Ngiri_23 Ngiri_25 Ngiri_26 Ngiri_27 Ngiri_28 Ngiri_29 Ngiri_30 Ngiri_32 Ngiri_33 Ngiri_34 Ngiri_35 Ngiri_36

St. Paul 2012 Severitya IRb 25 M 15 MSS 15 M 15 RMR 45 S 20 M 20 RMR 40 S 35 MS 30 M 20 M 15 M 0 R 20 RMR 15 MR 15 RMR 20 MS 25 RMR 40 MS 20 MR 25 M 20 MR 30 MS 15 RMR 15 RMR 25 M 10 MSS 0 M 20 MR 20 MR 35 MSS 20 RMR 35 MS 20 M 20 M 20 RMR 20 MR 0 R 30 MS

South Africa 2012 Severitya IRb 7 6 5 4 6 5 4 6 5 4 6 5 4 6 5 6 6 3 4 7 7 5 6 6 5 6 7 9 9 8 9 7 6 6 6 -

177

St. Paul 2013 Severitya IRb 15 MRMS 10 RMR 25 MRMS 25 MRMS 35 MRMS 15 RMR 20 RMR 40 MSS 35 MRMS 10 RMR 25 RMR 15 RMR 10 R 25 RMR 25 RMR 20 RMR 15 RMR 25 RMR 55 S 30 MRMS 25 MRMS 20 MRMS 40 MRMS 15 RMR 15 RMR 35 MRMS 10 MRMS 10 RMR 20 RMR 20 RMR 20 MRMS 10 MR 50 S 25 MRMS 15 RMR 35 MRMS 20 RMR 10 RMR 60 S

Kenya 2013 Severitya IRb 10 M 5 MR 5 M 1 R 5 S 5 M 5 M 30 MS 30 MSS 25 MSS 5 RMR 0 R 5 MSS 5 M 5 M 20 MSS 5 MSS 15 S 50 S 20 MSS 40 S 10 MSS 10 M 1 R 10 MS 30 MSS 1 R 5 MSS 5 R 5 M 35 S 5 MSS 15 MSS 20 S 5 MSS 15 M 30 MSS 25 MSS 30 S

Environment Line Ngiri_38 Ngiri_39 Ngiri_40 Ngiri_42 Ngiri_43 Ngiri_44 Ngiri_45 Ngiri_46 Ngiri_47 Ngiri_48 Ngiri_49 Ngiri_50 Ngiri_51 Ngiri_52 Ngiri_53 Ngiri_54 Ngiri_55 Ngiri_57 Ngiri_58 Ngiri_59 Ngiri_60 Ngiri_61 Ngiri_64 Paka_2 Paka_3 Paka_4 Paka_5 Paka_6 Paka_7 Paka_8 Paka_9 Paka_10 Paka_12 Paka_14 Paka_15 Paka_16 Paka_17 Paka_18 Paka_19

St. Paul 2012 Severitya IRb 35 MSS 35 MSS 30 MSS 0 R 50 S 10 MSS 15 M 40 S 40 S 15 M 20 M 25 MRMSS 15 RMR 25 M 20 RMR 15 MSS 30 MSS 20 M 35 MS 10 M 30 MSS 25 M 25 MRMSS 25 MRMSS 45 S 20 RMR 45 S 25 MRMSS 25 MSS 20 MR 20 MR 35 MS 25 MSS 25 M 25 MS 15 M 20 M 35 MS 25 MS

South Africa 2012 Severitya IRb 5 5 7 5 6 6 7 7 8 9 6 6 6 6 5 4 3 5 4 6 5 6 5 3 5 6 5 6 6 5 7 5 2 4 5 5 -

178

St. Paul 2013 Severitya IRb 35 MRMS 60 S 40 MRMS 5 R 35 MSS 5 R 10 RMR 65 S 55 S 25 RMR 20 RMR 30 RMR 20 RMR 20 MRMS 10 RMR 15 MSS 25 MSS 30 MRMS 15 RMR 40 MRMS 40 MRMS 30 RMR 10 MSS 25 MSS 30 RMR 60 S 25 RMR 25 MSS 15 MRMS 15 MRMS 35 MRMS 60 S 15 RMR 15 RMR 20 RMR 20 RMR 45 MSS 15 RMR

Kenya 2013 Severitya IRb 30 S 20 M 20 MSS 5 MSS 40 S 15 MSS 30 MSS 30 MSS 30 S 20 M 30 M 15 MSS 10 MSS 25 MSS 10 MSS 1 R 10 MSS 30 S 40 S 1 R 25 MS 30 MS 35 MSS 5 M 10 MSS 10 M 30 S 10 MSS 1 M 30 S 1 R 15 MSS 25 MS 20 M 10 MSS 5 MSS 15 M 20 M 15 M

Environment Line Paka_20 Paka_22 Paka_23 Paka_24 Paka_25 Paka_26 Paka_27 Paka_28 Paka_29 Paka_30 Paka_31 Paka_32 Paka_33 Paka_34 Paka_35 Paka_36 Paka_37 Paka_38 Paka_39 Paka_40 Paka_41 Paka_42 Paka_43 Paka_44 Paka_45 Paka_47 Paka_48 Paka_49 Paka_50 Paka_51 Paka_52 Paka_53 Paka_55 Paka_56 Paka_57 Paka_58 Paka_59 Paka_60 Paka_61

St. Paul 2012 Severitya IRb 15 M 20 M 15 MR 15 MR 20 MR 25 M 25 MRMSS 20 MR 25 M 25 M 25 MS 15 M 25 M 20 M 30 MSS 15 MS 15 MR 10 MR 10 MR 15 M 25 MSS 15 MR 20 MR 50 S 25 MRMSS 20 MR 10 MR 20 MR 15 M 40 MS 15 M 25 M 35 MSS 10 MR 20 M 40 S 30 MSS 20 M 20 M

South Africa 2012 Severitya IRb 4 6 5 6 5 4 5 5 6 6 9 7 7 6 7 7 6 5 6 6 6 5 5 4 4 5 4 6 5 5 5 6 5 -

179

St. Paul 2013 Severitya IRb 15 RMR 15 RMR 25 MRMS 20 MRMS 25 MRMS 20 RMR 10 RMR 10 RMR 60 S 35 MRMS 20 RMR 10 RMR 45 MRMS 15 RMR 35 MSS 15 MRMS 25 RMR 20 RMR 15 RMR 25 RMR 50 MSS 25 RMR 20 RMR 30 MSS 50 S 20 RMR 20 RMR 20 RMR 15 RMR 30 MRMS 10 RMR 45 MRMS 30 MSS 10 RMR 20 MRMS 25 MSS 15 MRMS 30 RMR 20 RMR

Kenya 2013 Severitya IRb 10 MR 5 MSS 1 R 5 MS 35 MS 50 S 25 S 15 M 20 MR 30 S 5 MS 20 MSS 20 M 20 M 30 MSS 5 MS 15 M 5 M 1 R 10 RMR 15 M 10 M 20 MSS 25 MSS 5 RMR 20 MSS 5 M 15 M 5 MSS 20 MS 5 MSS 35 S 1 RMR 1 S 10 MSS 20 M 10 MSS 5 M 5 M

Environment Line Paka_62 Paka_63 Paka_64 Paka_65 Paka_66 Paka_67 Paka_68 Paka_69 Paka_70 Paka_71 Paka_72 Paka_73 Paka_74 Paka_75 Paka_76 Paka_77 Paka_78 Paka_79 Paka_80 Paka_81 Paka_82 Paka_83 Paka_84 Paka_85 Paka_86 Paka_87 Paka_88 Paka_89 Paka_90 Paka_92 Paka_93 Paka_94 Paka_95 Paka_96 Paka_97 Paka_98 Paka_99 Paka_103 Paka_104

St. Paul 2012 Severitya IRb 25 M 25 M 40 S 20 M 20 M 20 M 20 M 15 MR 30 M 20 MR 15 MR 25 MR 0 R 35 MSS 25 M 35 MSS 20 M 15 MR 30 MSS 30 M 10 MSS 15 MR 35 M 25 M 20 M 15 MR 0 R 20 M 25 MSS 15 MR 35 MSS 25 MSS 20 MSS 45 S 25 MR 45 S 30 MSS 25 M 10 MR

South Africa 2012 Severitya IRb 5 6 5 6 7 6 5 5 4 5 5 4 4 3 6 7 4 5 3 4 5 5 5 4 3 5 4 3 5 5 5 8 5 3 -

180

St. Paul 2013 Severitya IRb 30 RMR 40 MRMS 40 MRMS 15 MS 35 RMR 40 MS 20 RMR 15 RMR 40 RMR 20 MSS 25 RMR 15 MRMS 5 RMR 35 MSS 40 MRMS 20 MRMS 50 MR 30 MRMS 25 MRMS 40 MRMS 40 RMR 20 RMR 40 MRMS 45 MSS 35 MRMS 20 MRMS 5 MSS 15 MSS 20 MRMS 10 RMR 30 MSS 25 MRMS 35 MRMS 30 MSS 30 MRMS 40 MRMS 55 MSS 20 MRMS 10 MRMS

Kenya 2013 Severitya IRb 15 M 20 MSS 20 MS 10 M 50 S 15 MSS 5 RMR 5 MS 30 S 10 M 10 MSS 5 M 5 M 10 M 25 M 25 MSS 15 MSS 10 MS 10 M 20 MS 20 MSS 5 M 30 MSS 35 MSS 20 MS 40 S 5 M 25 MSS 30 S 10 M 30 MSS 35 S 20 M 40 S 35 MSS 35 S 40 MSS 20 MSS 5 MSS

Environment Line Paka_105 Paka_107 Paka_108 Paka_109 Paka_110 Paka_111 Paka_112 Paka_114 Paka_115 Paka_116 Pasa_1 Pasa_2 Pasa_3 Pasa_4 Pasa_5 Pasa_6 Pasa_7 Pasa_8 Pasa_9 Pasa_10 Pasa_11 Pasa_12 Pasa_13 Pasa_15 Pasa_18 Pasa_19 Pasa_21 Pasa_22 Pasa_23 Pasa_24 Pasa_25 Pasa_26 Pasa_27 Pasa_28 Pasa_29 Pasa_30 Pasa_31 Pasa_32 Pasa_33

St. Paul 2012 Severitya IRb 30 MSS 40 S 40 S 35 M 40 S 30 MSS 20 M 25 M 30 S 25 MRMSS 15 MSS 5 R 15 MR 30 M 45 S 10 RMR 35 MS 10 RMR 25 M 50 S 35 S 15 MR 25 M 50 S 35 MSS 35 MS 15 MR 15 MR 25 MS 25 MS 30 MS 20 MS 20 MS 15 R 25 MS 25 M 15 M 15 M 20 MR

South Africa 2012 Severitya IRb 5 5 5 4 5 5 8 6 7 7 4 5 5 7 10 8 7 7 10 7 6 8 6 5 6 10 5 6 4 7 5 3 5 4 8 10 5 -

181

St. Paul 2013 Severitya IRb 40 MSS 50 MSS 45 MSS 35 MRMS 40 MSS 30 MRMS 10 RMR 15 RMR 35 MSS 25 RMR 10 MRMS 45 MSS 10 RMR 40 MSS 15 RMR 35 MSS 45 MSS 25 MSS 35 MRMS 15 MRMS 10 RMR 25 MSS 40 MSS 20 MRMS 15 MRMS 10 MRMS 30 MSS 15 MRMS 40 MSS 25 MRMS 25 MSS 25 MSS 15 RMR 15 RMR 25 MRMS 45 S 20 MRMS 45 S 10 MSS

Kenya 2013 Severitya IRb 35 S 40 S 25 MSS 40 S 25 S 30 MSS 25 M 15 M 25 S 20 MSS 1 S 20 MSS 5 M 5 MSS 10 M 20 MSS 15 M 5 M 10 M 5 MSS 1 R 10 M 20 M 1 R 5 MS 5 MSS 20 MSS 10 MSS 20 MSS 10 MSS 10 MSS 10 MSS 10 RMR 10 MS 15 M 15 MSS 50 S 20 MSS 10 MSS

Environment Line Pasa_34 Pasa_35 Pasa_37 Pasa_38 Pasa_39 Pasa_41 Pasa_42 Pasa_43 Pasa_44 Pasa_46 Pasa_48 Pasa_51 Pasa_52 Pasa_53 Pasa_54 Pasa_55 Pasa_56 Pasa_57 Pasa_58 Pasa_59 Pasa_60 Pasa_61 Pasa_62 Pasa_63 Pasa_64 Pasa_66 Pasa_67 Pasa_68 Pasa_69 Pasa_70 Pasa_71 Pasa_72 Pasa_73 Pasa_74 Pasa_75 Pasa_76 Pasa_77 Pasa_79 Pasa_80

St. Paul 2012 Severitya IRb 35 MSS 40 MS 15 RMR 10 RMR 20 MR 15 MR 15 R 35 MS 10 R 25 MSS 45 S 50 S 30 M 20 MR 10 R 10 R 15 MR 10 MR 15 RMRMS 15 MSS 35 MSS 55 S 25 M 10 MSS 30 MSS 25 M 30 MSS 15 M 10 MSS 15 MSS 45 S 5 MR 45 S 40 S 25 MSS 35 MSS 30 MS 20 MS 15 M

South Africa 2012 Severitya IRb 4 5 6 6 5 8 8 10 3 4 6 5 7 7 8 8 9 10 3 10 6 8 10 4 6 5 5 6 7 10 10 4 7 8 4 5 3 6 -

182

St. Paul 2013 Severitya IRb 15 MSS 20 MRMS 45 S 45 S 35 MRMS 25 MRMS 20 MSS 20 RMR 15 MRMS 15 RMR 30 MSS 20 MSS 15 RMR 25 MSS 20 RMR 15 RMR 20 MSS 15 S 15 MRMS 20 MRMS 10 RMR 30 MSS 30 MSS 15 S 15 RMR 35 S 15 MRMS 30 S 30 RMR 15 MRMS 35 RMR 40 MSS 35 MSS 20 MSS 40 S 15 RMR 25 MSS 15 RMR 15 RMR

Kenya 2013 Severitya IRb 1 MR 10 MSS 15 MSS 15 MSS 20 MSS 1 MR 1 R 10 M 1 R 1 R 10 M 10 M 5 M 10 M 25 MSS 5 MS 5 MS 1 M 5 M 20 MSS 5 MS 10 M 5 M 5 MS 15 MSS 15 MSS 1 R 1 MR 15 M 1 M 10 M 5 M 20 MSS 5 MSS 20 MSS 5 MSS 5 MSS 5 RMR 1 R

Environment Line Pasa_81 Pasa_82 Pasa_83 Pasa_84 Pasa_85 Pasa_86 Pasa_87 Pasa_88 Pasa_89 Pasa_90 Pasa_91 Pasa_92 Pasa_94 Pasa_95 Pasa_96 Pasa_97 Pasa_98 Pasa_99 Pasa_100 Pasa_101 Pasa_102 Pasa_103 Pasa_104 Pasa_105 Pasa_106 Popo_1 Popo_2 Popo_3 Popo_4 Popo_5 Popo_6 Popo_7 Popo_8 Popo_9 Popo_10 Popo_11 Popo_12 Popo_13 Popo_14

St. Paul 2012 Severitya IRb 40 S 10 M 35 S 40 S 15 MSS 20 M 15 MSS 15 M 15 M 40 S 40 S 15 M 10 RMR 15 R 10 R 15 MR 15 M 15 M 15 MS 30 M 15 M 20 M 35 MSS 15 M 25 M 30 MS 45 S 30 MSS 40 S 15 MSS 40 MSS 25 MRMSS 20 MS 20 M 20 M 25 M 15 MSS 50 S 25 MS

South Africa 2012 Severitya IRb 6 10 4 3 4 7 7 4 8 6 10 5 10 4 10 6 7 6 8 7 6 7 10 6 9 7 10 6 7 6 5 5 8 4 7 6 9 7 -

183

St. Paul 2013 Severitya IRb 20 MRMS 50 S 35 MRMS 15 RMR 25 MRMS 15 R 40 MRMS 45 S 30 MSS 35 MSS 10 RMR 15 MSS 20 S 25 MRMS 25 MRMS 40 S 15 MRMS 0 R 30 MSS 35 MSS 20 MSS 45 MRMS 10 RMR 15 RMR 25 MRMS 15 RMR 20 MRMS 10 RMR 15 RMR 25 S 10 RMR 15 RMR 15 MRMS 40 S 15 MSS 25 MSS 20 MSS 15 MSS 35 S

Kenya 2013 Severitya IRb 5 RMR 25 MSS 10 MSS 5 MSS 1 R 15 MSS 15 MSS 25 MSS 20 MSS 5 M 5 RMR 10 MSS 10 MSS 10 MSS 5 RMR 15 MSS 5 RMR 5 M 20 M 15 MSS 10 MSS 10 M 5 M 1 MS 40 S 5 MS 10 MSS 1 MS 5 MS 15 M 5 MSS 10 S 5 MSS 30 MSS 5 MS 5 M 10 MS 35 S 10 M

Environment Line Popo_15 Popo_16 Popo_17 Popo_18 Popo_19 Popo_20 Popo_21 Popo_22 Popo_23 Popo_25 Popo_26 Popo_27 Popo_28 Popo_29 Popo_30 Popo_31 Popo_32 Popo_33 Popo_34 Popo_35 Popo_36 Popo_39 Popo_40 Popo_41 Popo_42 Popo_43 Popo_44 Popo_45 Popo_46 Popo_47 Popo_48 Popo_49 Popo_50 Popo_51 Popo_52 Popo_53 Popo_54 Popo_56 Popo_57

St. Paul 2012 Severitya IRb 25 MSS 15 MSS 35 MSS 60 S 15 M 40 MSS 30 MSS 25 MSS 25 MSS 10 MS 30 MSS 30 S 25 MS 20 MS 45 MSS 40 S 15 M 30 MS 25 MSS 10 M 40 MSS 70 S 30 S 25 MSS 40 MSS 30 MS 15 M 25 MSS 30 MSS 20 M 20 M 10 M 35 MSS 40 S 20 M 40 S 30 S 40 S 20 M

South Africa 2012 Severitya IRb 10 7 5 5 5 5 8 8 6 8 3 5 5 3 5 5 6 2 6 8 6 5 5 5 7 6 2 5 5 5 4 6 8 5 8 4 -

184

St. Paul 2013 Severitya IRb 30 MSS 25 MSS 10 MRMS 25 MSS 15 MSS 20 RMR 20 MRMS 20 MSS 5 RMR 20 RMR 10 RMR 25 MRMS 15 MSS 30 MSS 45 MSS 20 MRMS 10 RMR 10 R 45 S 15 S 20 MSS 15 MRMS 20 MRMS 25 MSS 20 MSS 15 MRMS 15 MR 20 MS 35 MSS 20 MRMS 25 MRMS 30 MSS 25 S 10 RMR 50 S 30 MSS 45 S 40 MRMS 15 S

Kenya 2013 Severitya IRb 20 M 15 MSS 5 MSS 5 MSS 5 MS 5 MSS 18 MSS 10 MSS 20 MSS 20 MSS 5 MS 20 MSS 1 M 10 S 15 MSS 10 M 10 M 3 S 25 MSS 15 M 25 MSS 10 MS 10 M 10 M 10 S 15 MSS 5 MSS 10 M 10 MSS 25 MSS 5 RMR 10 MSS 15 MSS 1 R 20 M 15 MSS 40 S 10 RMR 5 MSS

Environment Line Popo_58 Popo_59 Popo_60 Popo_61 Popo_63 Popo_66 Popo_68 Popo_69 Popo_70 Popo_71 Popo_72 Popo_73 Popo_74 Popo_76 Popo_77 Popo_78 Popo_79 Popo_81 Popo_82 Popo_83 Popo_84 Popo_85 Popo_86 Popo_87 Popo_88 Popo_90 Popo_91 Popo_92 Popo_93 Popo_94 Popo_95 Popo_96 Popo_97 Popo_98 Popo_101 Popo_102 Popo_103 Popo_104 Popo_105

St. Paul 2012 Severitya IRb 70 S 35 MSS 30 M 40 S 25 MS 25 M 20 M 35 MSS 40 S 25 M 15 MS 25 MS 25 MS 35 S 20 MR 35 S 25 M 45 S 65 S 65 S 20 MSS 65 S 45 MSS 25 MSS 15 MR 25 MS 45 MSS 30 MS 30 MSS 20 M 35 MS 25 MSS 20 M 45 S 30 MS 35 S 30 MSS 25 MSS 40 S

South Africa 2012 Severitya IRb 3 6 5 7 4 4 4 6 6 8 5 7 8 6 10 4 3 4 5 4 5 5 3 4 6 4 4 6 8 7 3 6 5 6 6 5 6 -

185

St. Paul 2013 Severitya IRb 10 RMR 70 S 20 MRMS 50 S 10 MRMS 10 S 15 MRMS 15 MR 15 MRMS 30 MSS 20 S 35 MRMS 25 S 15 MRMS 25 MRMS 20 S 15 RMR 25 MR 20 MRMS 20 MRMS 15 MRMS 20 MSS 15 MRMS 15 MRMS 10 RMR 15 MRMS 10 RMR 15 RMR 50 S 25 MRMS 30 MRMS 30 MSS 15 MRMS 25 MSS 40 MSS 15 MSS 25 MRMS 35 MSS 15 RMR

Kenya 2013 Severitya IRb 10 M 10 MSS 10 M 25 S 5 M 5 MSS 10 M 10 MSS 15 M 30 S 25 MSS 30 S 10 M 10 M 10 M 10 MSS 1 MR 10 MSS 10 MSS 15 MSS 0 R 15 MSS 5 MSS 20 S 10 MSS 10 MSS 1 R 10 MSS 30 MSS 15 M 10 M 15 MSS 5 MSS 15 MSS 15 MSS 5 M 10 M 30 S 5 M

Environment Line Popo_106 Popo_108 Popo_109 Popo_110 Popo_111 Romany_1 Romany_2 Romany_4 Romany_5 Romany_6 Romany_7 Romany_8 Romany_9 Romany_10 Romany_12 Romany_13 Romany_14 Romany_15 Romany_16 Romany_18 Romany_19 Romany_20 Romany_21 Romany_22 Romany_23 Romany_24 Romany_25 Romany_26 Romany_27 Romany_28 Romany_29 Romany_30 Romany_31 Romany_34 Romany_35 Romany_37 Romany_38 Romany_39 Romany_40

St. Paul 2012 Severitya IRb 30 MSS 35 MSS 25 M 25 M 10 MR 25 MSS 20 M 20 MR 25 M 20 M 40 S 20 MR 20 MR 15 M 20 MR 25 M 30 M 30 MS 15 M 25 MSS 15 M 30 MSS 20 MS 25 M 0 R 45 MSS 35 MSS 25 M 15 MR 0 R 0 R 10 MSS 25 MSS 0 R 15 M 15 M 15 MSS 20 M 30 M

South Africa 2012 Severitya IRb 7 6 9 4 8 6 7 7 8 8 8 8 8 7 6 8 7 7 10 7 8 10 6 6 7 8 10 8 8 7 8 8 7 8 7 4 8 7 -

186

St. Paul 2013 Severitya IRb 30 MRMS 25 MRMS 10 MRMS 10 MRMS 10 MRMS 40 MRMS 20 RMR 5 R 40 MRMS 15 MRMS 30 S 35 MRMS 15 MRMS 40 MRMS 35 MRMS 20 RMR 15 RMR 30 MRMS 15 RMR 20 S 20 RMR 25 S 30 MRMS 15 RMR 0 R 65 S 35 S 20 RMR 15 RMR 0 R 5 R 15 S 55 MRMS 5 R 10 RMR 15 RMR 10 MR 30 MRMS 10 RMR

Kenya 2013 Severitya IRb 15 MSS 20 MSS 5 MS 1 RMR 5 MSS 5 MR 5 M 10 MSS 15 M 15 M 30 MSS 10 M 10 MSS 15 M 10 RMR 1 M 5 M 10 MSS 5 M 5 MS 5 MS 30 MSS 15 M 5 M 1 MR 30 S 25 MSS 25 MSS 5 M 1 S 5 MS 1 MS 10 M 1 MR 10 MSS 5 RMR 1 MR 15 M 5 M

Environment Line Romany_41 Romany_42 Romany_43 Romany_44 Romany_45 Romany_46 Romany_48 Romany_49 Romany_50 Romany_51 Romany_52 Romany_53 Romany_54 Romany_55 Romany_56 Romany_57 Romany_58 Romany_59 Romany_60 Romany_62 Romany_63 Romany_64 Romany_65 Romany_66 Romany_67 Romany_68 Romany_69 Romany_71 Romany_72 Romany_73 Romany_74 Romany_75 Romany_76 Romany_77 Romany_79 Romany_80 Romany_82 Romany_83 Romany_84

St. Paul 2012 Severitya IRb 25 M 35 MSS 35 MS 15 MSS 25 M 20 MR 15 MR 20 MR 40 S 35 MSS 50 S 25 M 40 S 40 S 25 M 45 S 25 M 15 M 20 MR 45 S 20 M 30 MR 15 MR 15 MSS 20 M 30 MSS 35 MSS 25 M 30 MRMSS 15 MR 20 M 20 M 15 MR 20 M 5 MSS 25 M 15 MR 20 MSS 20 M

South Africa 2012 Severitya IRb 10 9 8 9 9 7 8 8 8 8 8 2 7 8 10 9 8 8 10 7 7 7 10 7 10 10 7 2 7 4 6 7 10 5 10 10 10 6 -

187

St. Paul 2013 Severitya IRb 15 RMR 30 MRMS 30 RMR 20 MSS 15 RMR 20 RMR 20 RMR 25 MRMS 20 RMR 45 S 45 S 10 MSS 20 MRMS 40 MRMS 25 MRMS 40 MSS 30 RMR 20 RMR 15 RMR 60 S 35 RMR 35 RMR 20 RMR 5 MRMS 20 RMR 25 RMR 25 MSS 25 RMR 30 MSS 5 R 15 RMR 15 MRMS 15 RMR 30 RMR 10 MR 30 RMR 10 MR 10 MSS 10 MRMS

Kenya 2013 Severitya IRb 10 M 5 M 5 RMR 5 M 1 M 5 RMR 5 M 5 M 10 M 15 MSS 20 MSS 1 R 5 RMR 20 MS 5 MS 30 MSS 30 M 20 MSS 10 M 10 MSS 20 M 30 S 20 MSS 20 S 10 M 0 R 15 MSS 20 M 10 MS 5 MSS 10 MSS 5 MSS 15 MSS 10 M 10 MSS 40 S 10 S 5 MSS 5 MSS

Environment Line Romany_85 Romany_86 Romany_87 Romany_88 Romany_90 Romany_91 Romany_92 Romany_93 Romany_94 Romany_95 Romany_96 Romany_97 Romany_98 Romany_99 Romany_101 Romany_102 Romany_104 Romany_105 Romany_106 Romany_107 Romany_109 Romany_110 Romany_111 Romany_113 Romany_114 Romany_115 Romany_116 Romany_117 Romany_119 Romany_120 Romany_121 Romany_122 Romany_123 Romany_124 Romany_125 Romany_126

St. Paul 2012 Severitya IRb 60 S 25 MRMSS 30 MSS 15 MR 25 MSS 20 MSS 45 MSS 20 M 30 M 20 MR 25 M 20 MR 15 MR 25 M 20 MR 35 MSS 50 S 20 MR 45 S 20 M 45 MSS 20 MR 15 MR 30 MS 20 M 35 MSS 30 M 40 MSS 35 RMR 35 MSS 50 S 15 RMR 10 MR 5 MS 45 S 35 MSS

South Africa 2012 Severitya IRb 8 7 8 7 5 8 10 10 8 8 7 7 7 7 5 8 8 5 6 6 6 8 9 7 8 8 5 6 10 10 -

188

St. Paul 2013 Severitya IRb 70 S 15 MRMS 35 MSS 15 RMR 20 MRMS 35 MSS 20 RMR 10 RMR 30 MRMS 20 RMR 20 RMR 15 RMR 20 RMR 15 RMR 15 RMR 45 S 60 S 10 R 45 S 10 S 60 S 15 MRMS 20 RMR 25 S 20 RMR 40 MRMS 25 S 30 S 5 R 45 S 35 S 10 R 10 MRMS 10 R 40 MRMS 35 RMR

Kenya 2013 Severitya IRb 40 MSS 30 MSS 10 MS 10 M 5 MS 10 MS 40 MSS 10 MSS 50 MSS 30 M 25 MSS 40 S 40 S 10 MSS 5 M 15 MSS 15 M 10 MSS 10 M 40 S 40 MSS 10 M 10 MSS 15 S 10 M 40 MSS 10 S 15 M 10 M 5 RMR 5 M 5 MS 5 MS 10 MS 30 MSS 5 RMR

a

Disease severity (%) was measured according to modified Cobb scale (Peterson et al.

1948) b

Represents infection response of the plants to stem rust

189

Suggest Documents